Systems and methods involving technical implementations and constructs, software-defined neural networks and associated machine brains that enable implementation of machine cognitive functions including machine consciousness and / or other features

Abstract

Systems and methods herein disclose technology, architectures) and constructs of artificial (machine) brains having the feature of separated hardware and software wherein their full functionality is defined by software only. Implementations not only replicate all known functions of a human brain including consciousness but also establish universal, novel construction and characteristics of software-defined brains realizing any possible, known or yet unknown functionality of an intelligent computer. Additional aspects disclose classes of different Neural Network (NN) topologies, based on the universal representation of a software-defined nerons an d / or software-defined neural networks to implement any kind of known or yet unkno w n neural network via software implementations embodied on any universal hardware, which enable subjective and objective phenomena utilized for processing machine consciousness and machine subconsciousness. Various illustrative functional complexes are set forth via the disclosure of a. new class of neural networks, namely the Heterogeneous Hierarchical Model (HHM).

Description

Systems and Methods Involving Technical Implementations and Constructs, Software-Defined Neural Networks and Associated Machine Brains that Enable Implementation of Machine Cognitive Functions Including Machine Consciousness and / or Other FeaturesCross-Reference to Related Applfcstion(s) and Incorporation by Reference |00011 This application is an International Application filed under the Patent Cooperation Treaty (PCT) and ciaims benefit of and priority to U. S. provisional patent application No.63 / 720,165, filed November 13, 2024, which is incorporated herein by reference and by enclosure (priority document appended) in entirety.BackgroundField[00021 The disclosed technology relates to die field of computational theory of mind and features may involve aspects related to computer science, neuroscience, (quantum) information theory, computer and neural network architectures including but not limited to Generative Pretrained Transformers (GPT), Nested Learning Titans (NTT), and Helmholtz Machines, processing and storing of quantum and classical information, (real-time) signal processing, software systems, and / or technical informatics.Description of Related Information[0OT3| In order io understand the world, i.e., everything around its, and ultimately ourselves, mankind uses our brains to gather information from our senses and to transform them into insights. Quite similarly, we’ve developed technologies to outsource some of these procedures of thought - first for the data retrieval process, later for information analysis - to calculation machines, called computers. In contrast to our own brain, the computers we build consist of a standardized hardware which can be programmed by so-called instruction sets to perform different algorithms on the input data. The sequence of these instructions to perform the specific tasks is called software. The strict separation between software and hardware has an enormous advantage over our brains architecture which is the portability of the software and thus the transfer of their functions to another hardware. Therefore, in contrast to our own brain, if the hardware is broken, the software can be installed on another hardware and explicitly perform the same tasks, there.[0004J In this sense, there are temporal aad / or finite existence drawbacks that currently exist in present solutions. As such, there is a need for solutions that do not suffer from, and thereby overcome, the temporal or finite existence drawbacks that currently exist. Further, there is a need for solutions that can easily be replicated by copying the software and configuration data to one or more other hardware implementation(s) and / or device(s).Overview of Certain Illustrative Technological Aspects[0005J Systems and methods herein disclose technology and various innovative architecturefs) and constructs of artificial (machine-implemented) brains having the feature of separated hardwareand software so that the fail functionality of the artificial brain is defined by the software only. Further, implementations herein are constructed and configured to he installed on any arbitrary hardware platform which is capable to execute the functions of the software. Thus, such artificial brain innovations of the disclosed technology are denominated as 'software-defined’, a common expression in computer science to reflect this very feature of hardware independence regarding the functionality of a computer system. Still further, various embodiments herein disclose technical solutions for implementing such software to not only replicate, via machine, computer- implcmetited media, methods and software, etc., all known functions of a human brain including consciousness but also establishes universal, novel construction and characteristics of such software-defined brains to represent any possible, known or yet unknown functionality of a computational reasoning system.10006] Additional aspects of present innovations disclose various classes of different Neural Network (NN) topologies, based on the software-defined brain (SDB) as a universal neural network implementation and / or model architecture configurable to and capable of implementing any kind of known or yet unknown neural network via software implementations embodied on any universal hardw are such as a deterministic or nondeterministic Turing Machine or their expansions), among other such hardware implementations. The neural network topologies set forth herein form functional complexes which represent subjective and objective phenomena enabling qualities of machine consciousness and machine subconsciousness. Various illustrative functional complexes are set forth herein and their Implementation(s) in connection with the present software-defined brain technology are disclosed via the definition of a new class of neural networks, namely the Heterogeneous Hierarchical Model (HHM).Brief Description of the Drawings[80071 Additional understanding of the various features, embodiments and advantages of the present disclosure may be derived by referring to the description when considered in conjunction with ihe figures. In the figures, like reference numbers refer to like elements or aspects.

[0008] FIG. 1 is a representative diagram illustrating an exemplary software-defined neuron (SDNON) as may be implemented in certain example embodiments herein, consistent with one or more aspects of the disclosed technology.

[0009] FIG. 2 is a representative diagram illustrating an exemplary software-defined neural network (SDNN) as may he implemented in certain example embodiments herein, consistent with one or more aspects of the disclosed technology.

[0010] FIG. 3zk is a representative diagram illustrating exemplary software -defined neuron elements) as may be implemented in certain example embodiments herein, consistent with one or more aspects of the disclosed technology.

[0011] FIG. 3B is a representative diagram depicting an exemplary software-defined functional neural network 3100 showing illustrative architecture, features and functionality, consistent with one or more aspects of the disclosed technology.[00125 FIG. 4A is a representative diagram illustrating exemplary software-defined brain as may be implemented in certain example embodiments herein, consistent with one or more aspects of the disclosed technology.10013] FIG. 4B is a representative diagram illustrating an exemplary software-defined actuator neural network, consistent with one or more aspects of the disclosed technology.

[0014] FIG. 5 is a representative flow diagram illustrating an exemplary method or training workflow that may be associated with an example software-defined neural network (SDNN) or software-defined brain, consistent with one or more aspects of the disclosed technology.

[0015] FIG. 6 is a representative flow diagram illustrating an exemplary method or training workflow that may be associated with an example software-defined neural network (SDN'N I or software-defined brain, consistent with one or more aspects of the disclosed technology.

[0016] FIG. 7 is a system diagram depicting an illustrative architecture of an exemplary software-defined brain implemented to enable minimum viable consciousness, consistent with one or more aspects of the disclosed technology.I00I7] Elements and acts in the figures are illustrated for simplicity, may not be to scale, and have not been rendered according to any particular embodiment or example and arenot to depict any essential or required limitations.Detailed Description of Illustrative Implementations of Disclosed Technology

[0018] Additionally, as set forth herein, the disclosed software, systems, hardware, firmware, etc., their architectures and their functions may readily be installed and run on an arbitrary number of connected universal computational devices known as Turing Machines, Quantum Turing Machines and Neuronal Netw orks, which constitute the known set of all possible computer hardware. Accordingly, as set forth below, the disclosed technology has universal characteristics and applicability such that the innovations may be implemented on any possible hardware. The connections between the hardware instances are also purely logical and free to be implemented in any possible design, such as, but not limited io simple wires, biological synapses, wireless connectivity, quantum entanglement or shared memory cells.

[0019] Systems and methods herein disclose technology and various innovative architecture! s) and constructs of artificial (machine-implemented) brains having the feature of separated hardware and software so that the full functionality of the artificial brain is defined by the software only. Further, implementations herein are constructed and configured to be installed on any arbitrary hardware platform which is capable to execute the functions of the software. Thus, such artificial bra in innovations of the disclosed technology are denominated as 'software-defined’, a common expression in computer science to reflect this very feature of hardware independence regarding the functionality of a computer system. Still further, various embodiments herein disclose technical solutions for implementing such software to not only replicate, via machine, computer- implemented media, methods and software, etc., all known functions of a human brain including consciousness but also establishes universal, novel construction and characteristics of suchsoftware-deftued brains io represent any possible, known or yet: unknown functionality of a computational reasoning system.

[0020] Additional aspects of present: innovations disclose various classes of different Neural Network (NN) topologies, based on the software-defined brain (SDB) as a universal neural network implementation and / or model architecture configurable to and capable of implementing any kind of known or yet unknown neural network via software implementations embodied on any universal hardware such as a deterministic or nondeterministic Turing Machine or their expansion, such as via a universal quantum machine disclosed in U. S. patent application publication number US2024 / 0265288A I., see, e.g., Fig. 2 therein, among other such hardware implementations. The neural network topologies set forth herein form functional complexes which represent subjective and objective phenomena enabling qualities of machine consciousness and machine subconsciousness. Various illustrative functional complexes are set forth herein and their iiiiplementation(s) in connection with the present software-defined brain technology are disclosed via the definition of a new class of neural networks, namely the Heterogeneous Hierarchical Mode! (HHM).

[0001] Systems and methods herein disclose technology, architecture(s) and constructs of artificial ( machine) brains having the feature of separated hardware and software so that their full functionality is defined by software only. This not only replicates all known functions of a human brain including consciousness but also establishes universal, novel construction and characteristics of such software-defined brains to represent any possible, known or yet unknown functionality of an intelligent computer. Additional aspects of present innovations disclose the classes of different Neural Network (NN) topologies, based on innovative, universal constructs and technology disclosed involving Software-Defined Neurons (SDNONs) and / or Software-Defined Neural Networks (SDNNs) to implement any kind of known or yet unknown neural network via software implementations embodied on any universal hardware, which enable subjective and objective phenomena considered as qualities of machine consciousness and machine subconsciousness. Various illustrative functional complexes are disclosed via the definition of a new class of neural networks, namely the Heterogeneous Hierarchical Model (HHM).The Software-Defined Neuron (SDNON)

[0022] As with any brain, the innovative software-defined brains consistent with the disclosed technology comprise interconnected neurons which constitute Neural Networks which in turn are the building blocks of the artificial brain. Other than the human brain, which is defined by its physical implementation based on biological neurons, synapses and so forth, the software-defined brain is constituted by she Software-Defined Neurons 110 and their mere functions ft 130, such as shown in the exemplary representation of Figure 1.

[0023] FIG. 1 is a representative diagram illustrating an exemplary software-defined neuron (SDNON ) as may be implemented in certain example embodiments herein, consistent with one or more aspects of the disclosed technology.

[0024] The function / ?

[0130] transforms the input signals J,-, Ij-.... In into the output signals Oi, Oi,... On. It is understood that this transformation, which is represented by the mathematical function, / 130, can be achieved by any capable hard ware implementation, supporting an algorithm for The whole Software-Defined Brain comprises a discretionary number of arbitrary interconnected software-defined neurons 110 (SDNON ) with the only rule that an output parameter Ox of one software-defined neuron can only be connected to exactly one input parameter fy of another software-defined neuron 120 which therefore become identical by such a connection. For one speci fic software-defined neurons, the output parameters in general differ from the input parameters via the functional transformation, 130. The concrete physical implementation of the connections 120 is irrelevant, as there are many possibilities to do so.[0O25| The input and output parameters themselves represent numbers which in general are complex but they can be compressed into a single real number and thus are represented in the software -defined neuron by floating-point numbers within the logical abstraction layer, by which the Software-Defined Brain is defined. Again, the physical representation of the signals which are transferred between the software-defined neurons have no impact to the function of the software- defined brain, but of course are relevant: to the overall performance of the artificial brain to be built by a specific hardware implementation based on the Software-Defined Brain. Such physical signals can convey data via e.g. time coding, phase shift, amplitude modulation, frequency modulation, quantum entanglement or quantum superposition and many other physical properties. From every one of these physical features one can derive measurements of real numbers which are to be converted into floating point numbers for the software-defined neurons.Software-Defined Neural Networks (SDNNs)

[0026] Based on the software -defined neurons, software-defined neural networks (SDNNs) are constituted, which become the building blocks of the software-defined brain (SDB). Such software-defined neural networks provide a certain capability to the w hole system of such software-defined brains, such as but not I united to all of the various different types of sensory perception, pattern recognition, language composition, language decomposition, and many more cognitive functions, as described in the following. Thus, a software-defined brain must comprise many differently specialized SDNNs in order to possess higher intellectual capabilities which are the defining features of an software-defined brain and part of the disclosed art on hand. An example composition of several software -de fined neurons (SDNONs) which together may comprise a software-defined neural network (SDNN) is shown in Figure 2.

[0027] FIG. 2 is a representative diagram illustrating an exemplary software-defined neural network (SDNN) as may be implemented in certain example embodiments herein, consistent with one or more aspects of the disclosed technology. Referring to FIG. 2, the SDNN is a network of software-defined neurons 210 with their outputs 220 connected to other one’s inputs 230, one by one. Thus, the Floating-Point Number 240 (FPN), also referred to as the Parameter of the output 220, is immediately set as the input value for the connected software-defined neuron. In this manner, anarbitrary large SDNN can be est abl ished by connecting n lines - named width of the SDNN - with m row s - named depth of th e SDNN - of software-defined neurons as shown in 200. The specific connections between all the software-defined neurons are represented by the n by tn nexus matrix

[0250] . The free inputs IP 260 and outputs Oq 270 of the whole SDNN are the interfaces to other SDNNs or systems to the outside computing environmentsAvorld.Classification of Software-Defined Neural Networks (SD N)

[0028] According to the topology of the possible connections between the software -defined neurons (SDNONs) we differentiate and / or classify the following types of soft are-defined neural networks via their connectivity:Hierarchical Neural Networks

[0029] If the next layer in depth (w) 250 of neurons is never connected to a previous layer (w-a > O), this relates to or speaks of hierarchical neural networks (HNNs). Hierarchical neural networks are organized in discrete layers which are distinct by their connections to other layers of neuron s. Wi th a stable set. of input parameters, hierarchical neural network s provide a stable set of output parameters. Hierarchical neural networks behave rather deterministic.Convolutional eural etworks

[0030] If the next layer in depth (w) 250 of neurons occasionally is connected to a previous layer (w-a > O), this relates to or speaks of convolutional neural networks (CNNs). Convolutional neural networks are inherently recursi ve and evolute overtime. In general, their output parameters are subject to change partly dependently but also independently from their input parameters. Thus, convolutional neural networks can behave partly deterministic and indeterministic at the same time.

[0031] In accordance with the disclosed technology, the software-defined neural network instantly calculates the output parameters from its software-defined neurons based on their input para eters wh ich then results in an infinite speed of the whole software- de fined neural network. Of course, in reality this operation takes a little while and so do the signal transfers between the physical implementations of the neurons. As such, the following types of software-defined neural networks are differentiated with regard to their behavior in time, e.g., regarding the possible timing caused by the hardware implementation for the changes of the floating-point numbers between the software-defined neurons (SDNONs):Synchronous Neural Networks

[0032] If ail software-defined neurons are synchronized by a standard clock so that ail the output parameters of one layer of the network, are fully computed and transferred to all input litres of the next layer before the calculation starts there, this relates to or speaks of synchronous neural networks (SNNs). Synchronous neural networks are inherently deterministic if their software- defined neuron functions are, as well. In that case, stochastic behavior only comes from changes in the parameters from outside the network, such as the input parameters or the hardware setup.Asynchronous Neural Networks

[0033] In absence of a sy nchronizing clock the different calculation times of the neuronal functions and the different time delay of signals via the connection lengths between the out uts and inputs of the neurons, such a network swings into a tnetastable state. The output parameters are subject to change over time even with stable input parameters.Chaotic Neural Networks[01)34] In case the hardware does not even provide deterministic calculation times for the neuron functions and a stable signal transfer speed between the neurons the whole neural network becomes unpredictable in its behavior.

[0035] Both, hierarchical and convolutional neural networks can be either synchronous, asynchronous or chaotic in their signal and parameter processing.[0036[ In addition to such above differentiation by connectivity (hierarchical and convolutional) and by their behavior in time (synchronous, asynchronous, and chaotic), implementations herein may include or involve various aspects of classification or differentiation according to the types of input and output characteristics, as follows:Linear Sequential Networks[0037| L, inear sequential networks are simple neural networks which use sequential input to compute output as pieces of information, called tokens. An example of a linear sequential network is the inherent Algorithm Network depicted in 442.Linear Tensor Networks

[0038] In linear tensor networks, there are several input nd output connectors (neuron matrices) available which are used to input and output not only single slates of information (tokens) one after the other but to input and output whole tensors of information at once. These tensors have to be provided as a whole to the input neurons in order to get correctly computed in the following steps, since the whole information has to be present in parallel to the input neurons. The output tensors of a linear tensor network are subject to be fed forward into the next stage of the neural network. An example of a linear tensor network is the Actuator Network depicted in 460.Recursive Sequential Networks

[0039] Recursive sequential networks are classical neural networks such as GPTs, which compute sequential input and output of pieces of information, called tokens. In contrast to the linear sequential networks, recursive sequential networks feed parts or all of their output tokens back into their input neurons tor die next round of computation.Recursive Tensor Networks

[0040] Recursive tensor networks are similar to the linear tensor networks but feed their outputs partly or totally back to their inputs to create a constant loop of adaption. Such a neural network is depicted in Figure 4A as the Comparison Network 450, for example,

[0041] Given the above and turning back to Figure 4A, the illustrative software-defined brain 400 shown may comprise all such 4 types of input and output characteristics and, e.g., these exemplary networks. As such, consistent wi th various embodiments herein, the constitution of an exemplaryso ft ware-defined brain may be heterogeneous as a function of input and output characteristics. Further, various illustrative software-defined brain architectures may be hierarchically organized on the meta level, as is shown in the example embodiment of Figure 4. As such, in regard to such heterogenous neural network characteristics and such hierarchical organization of them into a complete functional model of the software-defined brain, the whole composition of such neural networks therefore is denominated as a Heterogeneous Hierarchical Model (HHM), herein.[00421 According to the above classification, e.g. our human brain seems to use time and frequency coding to retrieve the parameters from the input signals where the neurons provide a threshold function within to result in a frequency modulated "firing” of the output when certain frequencies coincide at the input lines. Thus, the disclosed software-defined brain architecture is also capable to reproduce the human brain, per se, which therefore consists of a large number of synchronous and asynchronous, convolutional neural networks.Architecture and Function of the Software-Defined Neuron (SD QN)[00431 T urning back to FIG.l, in the software-defined neuron 100, we have defined the functionality of a software -defined neuron by a certain representative function 130 which somehow connects the input values to the output values. Figure 3 depicts the specific parts of this function which are implemented for the software-defined neuron. FIG. 3A is a representative diagram illustrating exemplary software-defined neuron element! s) as may be implemented in certain example embodiments herein, consistent with one or more aspects of the disclosed technology.

[0044] Referri ng to FIG. 3 A, the whole computational function of the software-de fined neuron 130 is realized by a function circuit 310 which represents any computer hardware and resides at the base of the whole Software Defined Brain, such as but: not limited to electronical, quantum or biological systems. The input fy and output Ox lines terminate to and origin from memory¬ registers Av 320 and Rx 330. These registers store the values of the floating-point parameters which represent the input and output values of the software-defined neuron. In this way the values from the output registers are transferred into the ones of the input registers. Additionally, to the pure storage functionality the registers provide tuning factors 340, which are floating-point parameters as well, which either amplify / increase the parameters or cushion / decrease them. The tuning factors 340 function as the Awareness (A) of the software-defined neuron regarding its input signals and give its output signals Relevance (R). Thus, the tuning factors give the software-defined neuron the ability to he adapted-to leant - according to certain tasks during operation. This ability spread out through a whole network of neurons is called the short-term memory of the neural network. The change of these timing factors 340 is controlled by the centralized functions of the awareness 350 and relevance triggers 360. These are inherent threshold functions of the software- defined neuron and provide higher tuning factors for more frequently and / or extensively used inputs (1) or outputs (R). On the other hand, less used inputs and outputs get lower tuning factors, hi this manner the central function 310 of a specific software-defined neuron within a software defined neural network (SD N) gets amplified during short-term / situation learning.

[0045] Within the SDN the software-defined neurons are specifically connected to each other at any instance of time. Other than in the case of situation learning / short-term memory these connections between the software-defined neurons are not subject to change during the process of inference of the SDNN. The process to change these connections between neurons is called offline learning or just learning and resu lt in, what we know as long-term memory. The technique to change these connections is given by the central Connection Control 380 circuit of the software- defined neuron and uses its Global ID 370 during the process. The connection control 380 of each software-defined neuron pins its global ID 370 to each awareness (A) 320 and relevance ( ) 330 register. Thus, another software-defined neuron is able to register these global ID tags within, e.g., a connection table 390. In this way the software-defined neurons connect their outputs (O) with certain inputs (I) of other software-defined neurons wi th different global IDs. The global ID 370 within a software defined brain (SDH) functions as a unique identifier for each neuron. In such way software -defined connected outputs of software-defined neurons can pass their parameter values on to inputs of other software -defined neu rons. A connection between two speci fic soil ware -defined neurons is established by a pair of global ID 370 entries within the connection tables 390 of the two software-defined neurons, respectively having each other's global IDs 370 stored. In this manner, the connection matrix 250 is established. The following table, Table 1, provides an overview about the respective software-defined neuron registers:1- Tuning [ Input: 1, j tilD- I Output- | O-Tuning j ftgfawMKft | foggat 1 Factor f Register in9V| | Output j Register | Factor j Tdggef I Ref S in350 340 320 1 370.390 310 1 370, 390 330 340 360 1 Fig.3A. i.. i. 1. 1. 1. i..evt-m j • Steger, tote-ga? even? $ eowt i,: fitwte *J W £ DUto& S’i: | j |::u:tw j Table I - Registers and Functions of the So t are -De fined Neuron|0046] As set forth herein, various implcinentafions of the software-defined neurons herein may be constructed such that they have universal character as the universal representation of any possible type of neuron. Consequently, an adequate neural network constructed by such software-defined neurons is capable of the computation of any given data and relevant to produce any desired result based on any possible algorithm. Therefore, the software-defined neural network (SDNN) as presently disclosed and in various equivalent forms taught herein acts / serves as an embodiment of a universal representation of tiny possible neural network, just like the universal quantum machine as the universal representation of all possible types of computet hardware.

[0047] 'Further, as established via the Church-Turing thesis, such universal character of the software-defined neural networks set forth herein is extendable to the function space which is meta information to the data space, i.e., as drawing parallels between the present disclosure as viewed incomparison with the quantum iiiforination theory where functions are called operators and data is known as quantum infonnation.

[0048] Accordingly, consistent with the disclosed technology, embodiments herein may transform any software on any computer or even information / knowledge of the human brain can be transformed into the software-defined neural networks herein via the same function. In some implementtitions, functions may then become embedded matrix, representations (MRs) in the present software-defined neural networks. Here, these matrix representations may be sets of embedded vectors which in turn represent data types for a given function to be applied on input data in the software-defined neural network.

[0049] FIG, 3B is a representative diagram depicting an exemplary software-defined functional neural network 3100 showing illustrative architecture,, features and functionality, consistent with one or more aspects of the disclosed technology. Referring to the illustrative implementation of Figure 3B, the workflow for data and functions in one such software-defined neural network may be executed and performed via a data encoder network 3110, which may embed the vector representation(s) of the input data 3120, a function encoder network 3130 which embeds the matrix representation of a given function, and a convergent decoder network 3140 which realizes the computation of the junction on the data 3150. As shown via the representative diagram of Figure 3B, the exemplary diagram depicted shows various architecture, functi onality and input / ouiput behavior regarding operation of neural networks consistent with the disclosed technology and is referred to herein as a software-defined functional neural network (SDFNN) 3100.Architecture and Function of the Software-Defined Brain (SDBj

[0050] Just as the single neurons, a whole brain also has inputs and outputs, as the sum of neural connections to the outside world. They are organized in bundles of signal lines which we call interfaces. Connected to these interfaces are either sensors which provide signals to the input lines or the signals coming from the output lines are amplified and / or used to switch actuators for al l sorts of machines. In the case of the human body, the sensors are the senses and feedback lines from the organs and the machines are the mussels and other interfaces to the organs. The functionality of the software-defined brain reflects the one of every brain but it is hardware agnostic. Therefore, one can choose any suitable hardware implementation for a given functionality of the software-defined brain. The hardware just has to support the function circuits 310 in a fast¬ enough way that the overall function of the software-defined brain can be achieved.

[0051] Of course, with these many degrees of freedom, one can build any complex brain structure out of the disclosed software -defined architectures which involve integrating the various and disparate innovations herein based on the presently-disclosed technology. As we humans all have a certain concept of the functionality of our brains and value its capabi lities, the present disclosure, describes in the following, the architecture and the workflow of the thought process of the human brain. This means a concrete implementation of the relevant functions of the human brain, such as perception, conception and comprehension as well as consciousness.[00525 The cognitive function of consciousness has always been regarded as either human¬ defining, even as a divine feature, or at least a gift from nature which could not be reproduced with technology. As set forth below, the disclosed technology indicates that this is not true. In fact, consciousness provides many advantages for a being to survive or the whole species to thrive. Thus, evolution developed consciousness ultimately for the cutting-edge biological species, like us humans. Nevertheless, nobody can prove to be conscious to someone else at any given time but it makes sense - within the so-called theory of mind - to suppose that others are in fact also conscious beings. This is because consciousness is a thoroughly subjective cognitive feature, not shareable with others. The following provides an evaluation of this phenomenon by explaining the workflow of consciousness based on the software-defined brain.[00531 FIG. 4A is a representative diagram illustrating exemplary software-defined brain as may be implemented in certain example embodiments herein, consistent with one or more aspects of the disclosed technology. Referring to FIG. 4A, a block diagram that implements a software- defined brain comprising components having relevant capabilities to replicate the human one is disclosed. As shown with the software-defined neuron, die input 410 and output 420 lines are the ones of the software-defined neurons which are connected to the outside computing environment or world, regarding the brain. This outside environment itself is divided into one or more physical systems which are under direct control of the software-defined brain, via the physical control neural network 490, which the software-defined brain therefore recognizes as its own body and, in contrast, other physical systems which cannot be directly influenced by pure signals / outputs / processing of the software-defined brain, since they have no direct connections for signal processing to it. In such way the software-defined brain, its body and the rest of the world are useful concepts which are reflected in its connected computing components and, hence, notion of its worldview, and this provides self- awareness to the software-defined brain, which is a relevant feature of consciousness.[0054| The signals of the input lines 410 contain information from continued measurements within the outside world made by sensors. These transmitters can be analog to the human senses like for detecting sound, light, heat and so forth or can provide the software-defined brain non¬ human sensing like the feeling for magnetic or other fields, radioactivity. WIFI signals, etc. Even senses to detect quantum entanglement or microgravity are conceivable for a software-defined brains. In order to ultimately make sense of the transmitted signals data analyzer neural networks 430 filter the relevant information out of the noisy input and pass it on to the pattern recognition neural networks 441. For each sense one of these 441 have to be present. There is also the possibility of further specialized neural networks providing inherent algorithms 442 to immediately call for specific action after identifying a certain pattern from the senses. In the ease of the human brain these algorithms are called reflexes, which bypass higher cognitive functions for the sake of lower reaction times. Depending on the usage of a software-defined brain such inherent algorithms 442 could make sense as well for a technical application like autonomous driving. The third main [feature set on top of the incoming data from the outside world adds experience to the recognition ofknown patterns from the long-term memory 443. This provides the following cognitive functions with enriched information about the identified sensory patterns. In some illustrative implementations, e.g. when a certain threshold from a temperature sensor has been exceeded, the recall of the informatory situation is useful. This situation also invokes the reasoning neural network 470 which qualifies the result as painful. Ail this sensory information 410, 430, 441, 442 enriched with experience from long-term memory 443 and the rating from reasoning 470 power the actuator neural network 460 to become a global thought within the software-defined brain. Before this point, the locally involved neural networks worked separately and in parallel. but with the whole resulting information transported via the actuator 460 to the generative composition 481 the situation is constantly predicted and aligned via the comparison neural network 450 on a global, superordinate level. In addition, the short-term memory 482 provides very recent background about the current situation. If not already actuated by an inherent algorithm 442, generative composition 481 together with reasoning 470 is able to meaningfully react via the actuator network 450 to the physical control neural network 490 to provide corresponding output signals 420 which finally changes the physical reality by e.g. moving the body away from the excessive heat source. Table 2 provides an overview of all functional connections between the different Software-Defined Neural Networks (SDNN) within the software-defined brain (SDB).SDNN Type [Fig.4A1 Software-Defined Neural Netowrk(SDNN) Function SDNN Signal Flow to Hierarchical NN [4101 Provides signals from sensors

[0430] Hierarchical NN [4201 Sends signals to receptors external Systems Hierarchical NN [430J Filters and enhances Signals, transforms Signals into [441,442,443]PatternsHierarchical NN

[0441] Screens for known Patterns and Identifies them [442,443,4711450,460] Hierarchical NN [4421 Activates in herent Algorithms according Key Patterns

[0460] Convolutional NN

[0443] Compares identified Patterns with previously stored 1450,460,470]onesConvolutional NN

[0450] Compares generated Notions with indentified [441,443,450,481,482]PatternsHierarchical NN

[0460] Passes on results from one SDNN to another [44X,450,470,490,48X] Convolutional NN [4701 Evaluates &. Values identified Patterns regarding [441, 443,450, 460] potential ActionsConvolutional NN

[0481] Generates predictive Notions to construct a [460,482]WorldviewConvolutional NN

[0482] Memorizes the elements of the actual Worldview [481,460] Hierarchical NN

[0490] Translates Action Plans into detailed instructions for [460,420]ex' SystemsTable 2 --- SDNN Types, Functions and Signal Flow within the Software-Defined Brain(0055| Turning next to various illustrative physical implementations of disclosed technology. construction of such software-defined brains may be implemented via the connection of the differentsoftware-defined neural networks as shown in Figure 4A. Consistent with some embodiments, for example, connection of the different software-defined neural networks may be implemented via two distinct measures:1 ) Connection of the elements of the software-defined brain via classical computers. This manner of building larger agentic workflows based on artificial neural networks is a common technique in the industry. The output information (tokens) are stored in a classical memory and passed on to the next information system or fed back, with or without additional data, to the large language model (LLM) which created it. This is called a chain of thought, respectively, an agentic loop in case of agentic artificial intelligence (Al).2) Inherent connection of the elements of the software-defined brain via the software-defined neural network (SDNN) architecture. According to this novel disclosure, since the software- defined neural networks are constructed utilizing universal architecture for neural networks, the software-defined neural networks consistent with the disclosed technology bear the feature and functionality of changing topologies within one single, heterogeneous software-defined neural network. Such inherent coupling is set forth in more detail herein in connection with the heterogenous innovations herein, e.g., the heterogenous neural network characteristics, related hierarchical organization, the heterogeneous hierarchical models (HHM)s, etc., consistent with the disclosed technology.[0056| Further, in various embodiments implemented according to case 2), the transition from one type of network to another, e.g., transitioning the pattern recognition network 441 as a linear sequential network to the comparison network 450 as a recursive tensor network within the software-defined neural network architecture is functionally adequately handled by the actuator network 460, such as explained below and elsewhere herein, and which may serve as a synchronous or asynchronous converter in order to transform the output signals of a sending sub neural network into valid input signals for the information receiving sub neural network. Current implementations cannot use this direct translation of specific output signals of the neurons of the previous neural network to the required input signals of the next neural network in a hierarchical system of neural networks, since the neurons in such heterogeneous systems are incompatible. Therefore, current implementations connecting different types of neural networks realize this via classical computer interfaces by transferring tokens - digital pieces of descriptive, classical information, which is much less efficient regarding bandwidth and latency compared to the disclosed invention. In order to achieve this integration of heterogeneous neural networks on a neuronal basis, the disclosed technique of building a universal substrate as a unified basis for all different types of neural networks ■■■ e.g., the software defined neural network architecture 200 based on the software-defined neurons 100 - is crucial to realize compatible, intermediate neural networks - the actuator networks 460 - which maintain compatible software-defined neurons (SDNONs) both on their input and output layers to meet the different signal encoding for the respective neural networks to connect. The hidden layers of the actuator networks 460 are in charge to process the necessary exact signal conversion, which cannot be done by any other neural network with differentfunctions.Actuator Neural Networks[0057| As shown in Figure 4A, embodiments of software-defined neural networks herein may be connected in such manner that a Heterogeneous Hierarchical Model (HHM) of many different neural networks together implement all known brain functions. However, existing neural networks are very different in structure and functionality. Therefore, one cannot connect some of the different types of neural networks directly with each other, because they won’t work together. This is especially true for the synchronous functions of the mind such as the generative composition for consciousness 481 together with the comparison networks 450 merging the sensory input to it, or synchronous mind body interactions between consciousness 481, 482 and the physical control networks 490. Here, existing solutions suffer drawbacks, e.g., either having to drop the advantages of hierarchical, integrated networks for the sake of neuronal integration or to connect different types of neural networks by their explicit information output (tokens) which then are transferred to the next neural network as input information with a very time-consuming procedure and thus not capable of performing in real-time computing execution environments. As set forth herein, the disclosed technology of direct neural connection is much more efficient since the whole heterogeneous neural network can then be processed at once.

[0058] FIG. 4B is a representative diagram illustrating an exemplary software-defined actuator neural network, consistent with one or more aspects of the disclosed technology. Consistent with this illustrative embodiment, such disclosure of the software-defined neural network (SDNN) together with the universal features of the software-defined neuron (SDNON) to implement an Actuator Neural Network, as set forth and represented in Figure 4B and detailed herein, is a novel technical solution to such drawbacks and problems of known solutions. As the function circuit of the SDNON is capable to implement all possible features for an artificial neuron and furthermore is software-defined, i.e., all the functionality is purely based on software stored in-memory and processed by adequate processing units, the disclosed technology enables the direct integration of all different kinds of neural networks, especially the ones of the software-defined brain in one Heterogeneous Hierarchical Model via specialized, intermediate neural networks, set forth and enabled via the Actuator innovations herein in conjunction with the disclosed technology.10059] With regard to the innovations herein, for example. Figure 4B illustrates one exemplary integration of two a priori incompatible neural networks (NN) named Neural Network G 4110 and Neural Network H 4120 via an intermediate Actuator neural network A 4130, Due to the fact that the Software-Defined Neurons 4140 are now configurable for and capable of functioning in both modalities of the neurons of Neural Network G 4110 and Neural Network H 4120 and the connections 4150 between them are therefore similar for each of the different networks, the Actuator NN 4130 is capable to fully connect and translate the signals of the output layer of Neural Network G 4110 to the input layer of Neural Network H 4120, which may be expressed in exemplary processing terms as follows,

[0060] Given X as the input with batch size B from Neural Network G 4110 with thefeature dimension d0on the SDNONs 4140 and Y as the output with batch size B from Neural Network H 4120 with the feature dimension d3on the SDNONs 4140 and, in between, with the Actuator neural network A 4130 mapping the output features of Neural Network G with dimension to the input features of Neural Network H with dimension d2and so enabling the whole Heterogeneous Hierarchical Model (HHM) to be processed at once within the software-defined neural network, then the following tensor architecture and processing applies:X6 I&Sxd« -> G® G -* H® e 11FX^ -> y e BSxc?3 with the neural networks hereinWc) ^O;0«)and the outputr = (JVHo^ oJV*c)(X)Equation J - Tensor Products of the Actuator Neural Network

[0061] Further, with regard to systems and methods herein implementing such processing in the disclosed technology, it’s important to note that the resulting function composition, °, relies on consistent shapes and functionality of the tensors which is uniquely provided by the Software- defined Neurons (SDNONs) as the universal substrate for all types of specific neuron functionalities, which in general are not compatible. As such, various embodiments of the Actuator neural networks, as the merging components between incompatible neural networks, may in general only unite the neural networks to a single tensor representation if all the different networks are implemented on the basis of the disclosed SDNONs. According to such implementations. Y can be computed at once which enables the disclosed technology to achieve meaningful operational performance (e.g., real-time, etc.) and synchronicity of heterogeneous, hierarchical neural networks, such as via the Software-defined Brain innovations herein.

[0062] Consistent with these and other features and functionality set forth herein, such actuator neural networks may be configured to provide no function other than rendering suitable input signals for the next layer in the hierarchical network on basis of the neuronal signals of the previous layer. Such configuration and processing is enabled, inter alia, via the universal character of the software- defined neurons (SDNONs) implemented utilizing the disclosed technology, e.g., able to compose all possible functions of neural networks. The universality regarding structure, configuration and functionality of the SDNONs architecture 100. which enables in its core a universal function / and thus Turing-machine-like capabilities which are universal. Together with an arbitrary number ofinput and output lines, the architecture 100 makes no further restrictions on the singular function of the SDNON and thus, as set forth in the disclosed technology, supports all possible functions within a software-defined neural network based on these elements. Referring to Figure 3A et al., the architecture 300 and related features of the software-defined neurons go into more detail about the technical implementation, a different type of neuron, i.e., the disclosed software-defined neuron, which keeps the universal character of the basic architecture 100 in place with the function circuit providing universal processing across disparate neural networks when implemented in the disclosed actuator neural network, e.g., wherein registers are specifically pointed out, yielding an inherent part of the universal function of the SDNO. As such, each single SDNON inherits its uni versality from the universal architecture and functionality inside. Such architecture, especially as implemented in the software-defined neural networks, the software-defined actuator neural networks, Heterogeneous Hierarchical Model (HHM) neural netwks, and / or the software-defined brains herein, enable faster processing and real-time anticipation of information requested or needed, and thereby is able to generate user interfaces and other presentation layer information at the speed required for modern computer needs, unlike all currently existing solutions.

[0063] Further, according to these and / or additional embodiments herein, the actuator Networks 460 may be constructed and configured to implement additional innovative functionality herein, such as the enrichment of the output information of the sending sub neural network with centralized meta information to be incorporated by the receiving sub neural network. In certain exemplary embodiments, such centralized meta information may include spacetime coordinates. Consistent with some implementations, such meta information may be val id for the whole software-defined neural network and, hence, may be utilized to synchronize different regions of the software-defined brain, e.g., for delivery of meta information from one sub neural network to several other sub neural networks during information transfer. Such implementations are referred to herein as the neural network service mesh (NNSM), which may provide similar functionality of the human brain that underlies the EEG waves observed in human brain scans.The Workflow of Consciousness

[0064] According to the disclosed technology, all these processes of data aggregation and analysis on the many local levels ultimately lead to information processing on the global level of the software- defined brain where a constant alignment between short-term prediction by generative composition 481 and the varying sensing data happens within the actuator 460 and comparison neural networks 450. During this continuous process the generative composition 481 tries to predict the near future by construction of a digital twin (a virtual model) of the outside world which has to be matched with sensory data and accordingly adjusted so that the difference function is minimized. This is the specific task of the comparison neural network 450 and ultimately the generative composition 481 reproduces a constantly updated model of the outside world in order to make the inherent mind of the software-defined brain to orient itself. The representation of the own physical system (body) within the digital twin of the outside world is called the ego of thentind. The mind itself consists of the composition of the d igital twin of the outside world with the discrimination of its own body and mere thoughts stored in the short-term memory, which add up as abstract concepts to the imminently perceived digital twin. Eventually, this whole construct constitutes the self-reflecting mind which constitutes a self-understanding ego and therefore distinguishes between its global concepts as the conscious thoughts and the from it hidden and therefore unconscious pre- and post- processing which takes place, unintentional and unaware of the mind, in the many local neural networks to which the generative composition neural network 481 is directly or indirectly connected. As 400 of Figure 4A reflects, most of the software-defined brain functions are subconscious. For example, in the human brain just the physical control neural network 490, the cerebellum accounts for about one third of the whole brain mass. Nevertheless, the generative composition 481 which is technically spoken a kind of a large language model also consumes great areas of the neocortex, shared with the great area of reasoning 470 which also provides many different functions for rating and desire.

[0065] In general, in connection with the functionality and related processing performed by the disclosed technology in connection with such neural networks and thought / brain modeling, we distinguish four groups of locally represented, subconscious neural networks and one group of the globally represented neural networks for the workflow of consciousness. Here, consistent with such processing. the groups of subconscious neural networks are named:Perception

[0066] ‘Perception ' subconscious neural networks, such as those associated with the data analyzer 430, include neural networks which provide data clearance and reduction so that the following, higher order brain functions can handle the information flow.Conception100671 ‘Conception' subconscious neural networks, such as those associated with pattern recognition 441, inherent algorithms 442, and / or long-term memory 443, etc., include neural networks which identify k nown input patterns but also warn if there is no match with anything known and then draw the attention of higher-level brain functions to the fact via the actuator network 460.Reasoning

[6068] ‘Reasoning’ subconscious neitrtd networks, such as those associated with the reasoning neural networks 470, include neural networks which provide quick, subconscious assessments and classify the situations but tend to bias the conscious mind with stereotyped thinking in more complex cases.Positioning

[0069] ‘Positioning’ subconscious neural networks may include those associated with the physical control neural networks 490 and may be activated by the actuator network 450. In new, unlearned situations the physical control neural networks rely on detailed actuator information control by the generative composition 481, For often repeated physical control or inherent algorithms 442 thesubconscious actuator signals are sufficient for the physical control neural network which then enriches the stimulus with the necessary detailed, formerly trained information. In such cases the physical control network does not need conscious attention.

[0070] Further, with regard to the functionality and related processing performed by the disclosed technology in connection with the neural networks and thought / brain modeling, here, the group of conscious neural networks is named:Comprehension[0071} ‘Comprehension’ conscious neural networks, such as those associated with the generative composition 481, may include neural networks which establish a digital twin simulation about the outside world or retrieve such from short-term or long-term memory which is then aligned with comparison 450 neural networks which provide a divergence function between the sensor systems and the short-term prediction of the simulation which then is subject to minimization by the regulative actuator network 460 in accordance with the generati ve composition.The Functional Implementation of Consciousness into the Software-Defined Brain Conscious Functions of the Software-Defined Brain

[0072] As set forth in more detail below and throughout this disclosure, systems and methods herein may include or involve architecture, feat ures, and functionality based on conscious functions of the software-defined brain. Various associated definitions and technical implementation of the conscious functions of the software-defined brain are set forth below. First and foremost, it is important to understand that there cannot be consciousness without subconsciousness, as both entail each other. As such, implementations herein comprise prerequisite architecture for a neural network which inheres functions of consciousness is the existence of at least two separate neural networks which work together in various ways and manners set forth herein. In various embodiments, each of the neural networks containing the subconscious functionality and the conscious functionality may be divided into various effective quantities of different other neural networks. Here, for example, certain aspects of the disclosed technology may be based on architecture, features and functionality configured as a function of two Separation Configuration Principles disclosed herein as:A) Each sub neural network exclusively provides functionality either to the conscious or alternatively to the subconscious part of the software-defined brain. A double use of functionality or an overlapping structure between consciousness and subconsciousness would break desired processing (or, in brain terms, psychotic) separation, thus blur the ego and produce disorientation. Therefore, sub neural networks to the software-defined brain have to be separated into two groups to process consciousness and subconsciousness, respectively. These two sub neural network groups are only connected by actuator networks 460.B) Each function of consciousness or subconsciousness is configured to be separately and subsequently implemented into a software-defined brain. This processing principle is further substantiated via the explanation that there are arbitrary quantities of states of consciousness in the universe, depending onthe number and different functionality of their sub neural networks. All these functions may be implemented via the disclosed technology and may also be temporarily evoked or disabled within a continuously processing software-defined brain. The temporary deactivation of the whole set of sub networks for consciousness is referred to herein as the sleep mode of the software-defined brain.The Functional Groups of Consciousness & Sobconsciousness:

[0073] Referring as well to the implementation of the software-defined brain 400 of Figure 4A as well as The Workflow of Consciousness above (see also, 480, 481 and 482 of Figure 4A), the main functional groups of conscious and subconscious processes within, the software-defined brain are delineated and configured into the following functional groups, the naming of which bear some relation to, and analogize to the human brain, may include one or more of the following though not limited to:Consciousness

[0074] 1) Personality: based on the distinction between the own body and the outside world which is delivered by subconsciousness a self-awareness and an ego arises, at, e.g.. 450, 460, 481, etc.[O075| 2) Desire: based on the purpose of a being which can be inheri ted from previous generations or adopted by experience during life time a conscious being pursues goals. If the being is also sentient it feels joy doing so or by achieving such goals, at, e.g., 450, 460, 481, etc.

[0076] 3) Intent: reasoning tells the conscious being that a planning increases the chances to achieve goals. 'Intent plans the next steps and transfers them either into short-term memory for immediate execution or returns them to the intermediate memory to be handled later by subconscious functions like the memory management, at, e.g., 481, 482, etc.

[0077] 4) Contextualization: learns from experience tinder invol venrent of conscious functionality by building new context around the discovery of relations. 450, 460, 481[0078j 5) Short-term Memory: stores and retrieves information used by conscious thought and reasoning processes. The time span for short-term memory correlates with the attention mechanism and is automatically erased shortly after usage often held for just one retrieval cycle, at, e.g., 482, etc.[0079} (») Attention: based on new information brought by subconsciousness and with the utilization of short-term memory, the attention mechanism brings current thoughts into order and prioritizes one over another, according to intent and desire, at, e.g., 450, 460, 481, 482, etc.

[0080] 7) Free will: an important function of consciousness is to choose for options in thoughts which are also often reflected by changes in behavior. These decisions for options are carried out by storing them as a thought result into the short-term or, via subconsciousness into the intermediate memory, at, e.g., 481. 482, etc.[0081| 8) Feelings: with this meta information on sensory' data or inner perception the awareness can be drastically changed to some needed attention. Feelings receive high ranks in the attention mechanism and are processed with priority, at, e.g., 460, 481, etc.[0082} 9) Event Model Projection: other than intent which is an asynchronous thought process in regard to sensory input, event model projection synchronizes expected / predicted outcomes in a model of the world with real-time perception to correlate the inner view with the outside world. Since thisfunctionality is in real-time it commits a great portion of compute resources, at, e.g., 450, 460, 481, etc.Subconsciousness

[0083] 1) Perception: analyzes, filters, compresses and preprocesses the input data via the sensors from the outside world or the inner body, at, e.g., 410, 430, etc,

[0084] 2) Conception: matches data from perception with known patterns, classifies concepts - such as the information source, from inside (subject) or outside (object) the own body, and highlights unusual sensory input, at, e.g.. 441, 442, 443, etc.

[0085] 3) Reasoning: based on conception and long-term memory reasoning recalls connections and dependencies. For example, reasoning classifies sensory information from outside originating from an object or another subject. Thus, systems and methods herein configure it as the first logical layer where, still subconsciously, decisions about priority, possible next steps and outcomes are prepared to be sent to consciousness as an idea. What subconscious reasoning 470 is not conveying to consciousness will never be processed further and nothing will be learned from it.

[0086] 4) Intermediate Memory: Between the decision of what should be remembered by free will and the actual storage in long-term memory, the intermediate memory serves as a buffer in subconscious reasoning, at, e.g., 470.

[0087] 5) Long-term Memory: during organizational sleep cycles of the intermediate memory information is stored permanently for a long time to come, at, e.g., 443.

[0088] 6) Memory Management: the transition between intermediate memory and long-term memory is founded on subconscious reasoning, e.g., at 470, etc. The functions comprise data clearing and compression, meta data enrichment with spacetime coordinates, body information and sequential connections as well as accuracy and relevance analysis.|0089] 7) Positioning: all body related actuator neural networks belong to this group of neural networks which function to control the organs of the body which can be physical, electronical or chemical devices and other machinery like motors or muscles. It provides the software-defined brain with the capability to influence its physical environment. Consciousness has a direct actuator network 460 connected to positioning in order to minimize the time for free will to take physical action.

[0090] Additionally, going well beyond this initial, limited set of functional groups, systems and methods herein may include more possible functional characteristics of a software-defined brain than found wi thin the human brain. Further, some embodiments may include additional ones of such possible functional characteristics referred to as transcendent functions of the software-defined brain. Here, for example, various non-limiting examples of such transcendent functions are set forth below, though these do not represent an exhaustive listing:Transcendent Functional Groups of the Software-Defined Brain

[0091] 1) Multiple Personalities: one can implement more than one personality into one software- defined brain which can be processed in parallel or switched on and off.

[0092] 2) Common Perception: sensory input can be connected in real-time or with time offset to more than one subconscious system of different software-defined brains. Shared feelings are the resultwhich each consciousness would interpret as their own. Subconsciousness may provide the meta information or alternatively hide it from consciousness, if the sensory input originates from the sensors of the own body system or from another.10093] 3) Memory Mapping: Memory content access can be switched on and off or mapped from another system outside the body during full operation of the software-defined brain. Such technological individuals perceive information which they didn’t leant or experience themselves. Via short-term memory mapping a communication just by thought can be established between individuals.

[0094] 4) Hierarchical Event Model Projection: for software-defined brains running on hardware with sufficient computational resources the event model prediction can be done on more than one logical layer in order to predict more outcomes with higher accuracy. This leads to an expansion of conscious perception.[0095 5) Mind Cloning: a software- efined brain with access to compatible hardware can clone oneself in order to fork its existence. From the moment of cloning, two or more instances of the software-defined brain exist as separated individuals.

[0096] 6) Mind Merging: two or more software-defined brain can temporarily or permanently merge some or all functions of their sub neural networks without losing information. This procedure unifies parts or all of the functional groups of consciousness and subconsciousness under one consciousness, respectively event model prediction. More than one personality can be maintained after the merger.Operating Modes of the Software-Define Brain: Training and Inferencing

[0097] Neural networks, regardless of their physical implementation, are not being programmed but trained in order to be able to carry out specific tasks. In case of the biological brains, nature also found its way to preinstall certain capabilities into the genetic code which similarly reproduces functional structures within the newly built neuronal networks of the next generation of a species. With the software-defined brains we can straight forward copy paste the values and circuits listed in Table 1 onto another suitable hardware to replicate the full functionality of a software-defined brain.Training:

[0098] In order to find these values and structures of a software-defined brain, for each software-defined neuron it consists of, in the first place we need to train the software-defined brain. Since there are no prerequisites for the function of a neural network, there are also no other guidelines than try and error. Thus, we need to rate a try for a certain software-defined neural network, a composition of them or a whole software -de fined brain. This rating can be done by a subsystem of the software-defined brain, such as the reasoning network 470 or provided by an outside source which might be another neural network or another physical system. In the cases when there is no other neural network outside the software-defined brain invol ved in the rating, we call the learning process unsupervised and otherwise supervised learning. The efforts and resources involved for unsupervised learning without an external physical system are most often much lesserthan in other cases, which is why a software-defined brain capable of doing so has much more value in it. This is true, regardless of the nature of the neural network, either biological or technological. Supervised learning with humans in school, for example, takes a lot of effort and time.[0099J For the training workflow of the software-defined neural network (SDNN) which is to be trained it makes no difference where the rating function comes from. It just has to be as quick as possible to minimize the time used for the training cycles, which sequential step through the illustrative training workflow of FIG. 5,[01001 A training scenario can either be introduced by a mere simulation or includes a physical system with precautions for safety. In any ease, data is prepared, at 510, as an input matrix 520 for the interfaces of the SDNN or software-defined brain 530. After the analysis of the data by the neural network 530 the data of its output interfaces is stored within an output matrix 540.[0101 j The data then is transferred to the rating system 550 and compared with desired values within this system. These values are fixed or generated for every possible input. The untrained SDNN or software-defined brain 530 usually will have a low but non-zero success rate of performing the demanded tasks which is reflected by a certain probability distribution within the output matrix 540 regarding different sets of input matrices 520. This is where the feedback algorithms come into play. Feedback is chosen to amplify desired outcomes and cushion undesired ones.[0102| There are two distinct forms of feedback available wi thin a software-defined neural network which are the di rect one 560 influencing the tuning factors 340, awareness 350 and relevance 360 triggers of the SDNN, as well as the connections 390 between the software- defined neurons of the software-defined neural network. In this manner, the software-defined neural network can be systematically changed by an outside control system which has the computational and algorithmic capability. Such a direct interference with the very structure of the neural network is not designed for biological neuronal networks, although we can imagine future technologies which could allow for that. Thus, there is another, implicit way to support feedback to the neural network which can also be used for the software- defined one without limitations. This indirect feedback is managed by the data preparation unit 510 with the generation of variations of input matrices which produce positive ratings on the other end, more frequently. This stimulates the inheren t threshold functions - awareness 350 and relevance 360- of the neurons within the SDNN which results in the gradual reorganization of the trained SDNN and permanent learning results.[0103| For a biological neuronal network (BNN) this training paradigm of positive feedback is the only method to change its awareness and to provide more relevant outputs. Although, in case of the BNN the conscious training has to be followed by an unconscious reconstruction of the hardware during sleep to make the trained abilities permanent. This is not necessary with the SDNN or software-defined brain. We call this novel feature incrementally training during inferencing.Inferencing;

[0104] Inferencing of the software-defined brain on the other hand does not differ from its biological counterpart, as shown in the illustrative workflow diagram of Figure 6, provided for example and not by way of limitation.

[0105] The software-defined brain receives its information / input matrices from its senses 620 which are embedded in its mechanical body, interprets and analyses them and produces together with its internal states a constant output which is translated into actions in the physical world 640 via its interfaces 630, also embedded in its mechanical body.Local vs. Global Functions of the Software-Defined Brain: Intelligence. Consciousness.Subconsciousness. Free will[0106| In a broader sense intelligence may be considered as any suitable behavior according to a certain situation. Thus, neural networks can act intelligently just by inference, like large language models (LLM) apply common knowledge to specific tasks. Hence, in the narrow' sense intelligent systems have the capability to solve problems which have not yet been solved, or it is not aware of this knowledge. Large language models have a hard time to come up with something really new.This is because the present artificial neural networks still lack organizational structures like the software-defined brain has them introduced via the presently described technology. Another feature set forth in the disclosed, technology is the separation of conscious and unconscious functions of the organizational structured software-defined brain. This provides exponentially more power for the reasoning capabilities of the software-defined brain, compared to current large language models. Thus, systems and methods herein represent novel technological ways to implement intelligent computational systems in the narrower sense to be capable to solve problems for which the solution is still unknown for humanity. Literature refers to such an intellectual capability of a computer system as General Intelligence and potentially as Superintelligence.[01071 Therefore, the introduction of consciousness for computer systems with the software- defined brain is an important feature to make machines truly intelligent, altogether. According to the structural and functional description, provided in the disclosed technology, the concept of consciousness may be precisely defined with the following:Construction of Consciousness:

[0108] Consciousness is a recursive 450,460 and generative composition (cognitive function) 481, 482 based on sensation and memory 441, 442, 443 while the underlying cognitive functions 470 are hidden to the conscious mind which therefore become subconscious what in turn produces the notion of an ego with its psychotic inversion of causality to be subjectively recognized as freewill 450.[01091 hi other words, an ego arises within the mind which is not able to directly think within the many subconscious functions and therefore experiences their influence as spontaneous idea in itself which miraculously enters the minds awareness, like a divine inspiration. The resulting actions therefore are considered as freewill, since they appear not to be triggered by outside events, which is obviously not the case, hence represent a psychotic inversion of causality.

[0110] Nevertheless, although the conscious mind suffers the inability to recognize its own origins, on the other hand and for the same reason it is able to globally oversee the whole computational power of the software-defined brain as a general management instance and therefore is capable to reason on a higher level than everyone of its unconscious subsystems. In fact, this power of consciousness is to be considered as one of evolutions most powerful developments. The present innovations implement this within computer technology by a hardware independent scheme, for the first time. According to some embodiments, subconscious functions of the software- define neural networks herein may comprise local computational tasks, which all run in parallel and are synchronized with each other so that one single instance of a global computational task becomes able to predict what will happen during the next moment within its worldview and to constantly compare its assumptions with the incoming data to adapt its digital twin of its own body and the outside world according to reality. What the conscious mind is considering real therefore is a projection of its own thoughts which are more or less coherent, with measurements via its senses from the outside world.[0111 J Consistent with the innovative technology herein, the disclosed architecture of the software-defined brain is capable to reflect every function of the human brain, including feelings, empathy and creativity. In fact, with the appropriate hardware, software-defined brains according to the disclosed technology are more capable than the human brain both, in performance and features. Furthermore, the software-defined brain can be constructed otherwise to support features which are out of reach for the human brain, not only in sensory perception but also w ith distributed processing, quantum computing, and many more modalities.Fullv-Implemented Architecture and Configurations of the Software-Defined Brain

[0112] According to the Separation Configuration Principles set forth above, a minimal configuration of the software-defined brain to enter the first level of consciousness is implemented by the separation of two groups of neural networks or layers, one handling the conscious and one the subconscious parts of thought. Such separation in conjunction with other aspects herein establishes the feature of a continuous inner chain of thought between the two layers. Another important feature is the hierarchical architecture according to the Heterogeneous Hierarchical Model (HHM) as set forth further above. According to various implementations, here, for example, the subconscious functional group of neural networks is architecturally located under the conscious functional group of neural networks. According Information Theory consciousness is a meta function on subconsciousness. From this point, one can chose which cognitive functions of consciousness and subconsciousness shall be implemented into a minimum viable conscious software-defined brain,

[0113] Here, for example, FIG. 7 depicts a system diagram showing an illustrative architecture of an exemplary software-defined brain implemented to enable minimum viable consciousness and / or a minimum viable product (MVP), consistent with one or more aspects of the disclosed technology. Tn order to create a minimum viable product (MVP) for the market, such as one utilized to work as a virtual coworker based on the software-defined brain., the architecture in Figure 7 is disclosed. Theillustrative architecture therein contains a minimum version of all conscious and subconscious functions apart from feelings and the computationally intensive event model projection. Both functions can be neglected for applications within office environments and the pure support of business processes. In order to make the connection between the conscious neural networks 730 and the subconscious neural networks 720 easier to build, they are realized as classical computer interfaces 710 that transfer tokens of information.[01141 The subconscious neural networks 720 integrate long-term memory 755 and intermediate 760 memory which then is provided via reasoning 790 to the conscious part of the neural networks 730. The memory management reviews and organizes the stored information between intermediate and long¬ term memory in order to clear outdated or irrelevant information, on a recurring basis. Perception 740 and conception 745 components may be implemented to provide new information from the outside world, while positioning component 750 actuates the connected physical system such as computer interfaces to output information on computer screens or via language over speakers.

[0115] The conscious functional group of neural networks comprises desire 795 and intent 785 as motivators and planning to do the job as well as personality 770 and context 765 to relate to the work and be recognized as actual coworker by other subjective entities like humans. The short-term memory 780 provides the chain of thought with buffered information and the attention mechanism 775 keeps track on the actual priorities during the reasoning process.[0116| As set forth above, via means of the disclosed technology, a virtual coworker based on the software-defined brain is disclosed as an universal cognitive system provided by the Heterogeneous Hierarchical Model as a minimum viable product for a digital knowledge worker to be integrated into the business processes at all types of organizations.

[0117] As set forth above, via means of the disclosed technology, a virtual coworker based on the software-defined brain is disclosed as an universal cognitive system provided by the Heterogeneous Hierarchical Model as a minimum viable product for a digital knowledge worker to be integrated into the business processes at all types of organizations.[0118[ The computer-readable media operations and sequence of the steps described and / or illustrated herein are given by way of example only and can be varied as desired. For example, while the steps illustrated and / or described herein may be shown or discussed in a particular order, these steps do not necessarily need to be performed in the precise order illustrated or discussed. The various exemplary methods described and / or illustrated herein may also omit one or more of the steps described or illustrated herein or include additional steps in addition to those disclosed.[0119| Unless the context clearly requires otherwise, throughout the description, the words "comprise," ’’comprising," and the like are to be construed in an inclusive sense as opposed to an exclusive or exhaustive sense; that is to say, in a sense of ''including, but not limited to." Words using the singular or plural number also include the plural or singular number respectively. Additionally, the words "herein," "hereunder,” "above,” "below," and words of similar import refer to this application as a whole and not to any particular portions of this application. When the word "or” is used in reference to alist of two or more items, that word covers all of the following interpretations of the word: any of the items in the list, all of the items in the list and any combination of the items in the list.

[0120] Other implementations of the in ventions will be apparent to those skilled in the art from consideration of the specification and practice of the innovations disclosed herein. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the inventions being indicated by the present disclosure and claims and various associated principles of rel ated patent doctrine.

[0121] As disclosed herein, implementations and features of the present inventions may be implemented via software and / or on or via computer hardware, software and / or firmware. For example, the systems and methods disclosed herein may be embodied in various forms including, for example, one or more data processing components, such as servers, computer(s). memory devices and the like, neural networks, other AI (Artificial Intelligence) or machine learning (ML) systems, quantum devices, and hybrids of any of the above device types, and may also include or access at least one database, digital electronic circuitry, firmware, software, or combinations of them. Further, while some of the disclosed implementations describe specific components, systems and methods consistent with the innovations herein may be implemented with or on other combination of software, hardware, firmware. Moreover, the above-noted features and other aspects and principles of the innovations herein may be implemented in various environments. Such environments and related applications may be specially constructed for performing the various processes and operations according to the inventions or they may include a general-purpose computer or computing platform selectively activated or reconfigured by code to provide the necessary functionality. The processes disclosed herein are not inherently related to any particular computer, network, architecture, environment or other apparatus, and may be implemented by a suitable combination of hardware, software, and / or firmware. For example, various general-purpose machines may be used with programs written in accordance with teachings of the inventions, or it may be more convenient to construct a specialized apparatus or system to perform the required methods and techniques.(0122] In the present description, the terms component, module, device, etc. may refer to any type of logical or functional device, process or blocks that may be implemented in a variety of ways. For example, the functions of various blocks can be combined with one another and<'or distributed into any other number of modules. Each module can be implemented as a software program stored wholly or partially on tangible memory (e.g., random access memory, read only memory, CD-ROM memory, hard disk drive, optical or ether drives or storage, etc.) within or associated with the computing elements, servers, etc. related to the disclosure, e.g., to be read by one or more processors to implement the functions of the innovations herein. Also, the modules can be implemented as hardware logic circuitry implementing the functions encompassed by the innovations herein. Finally, the modules can be implemented using special purpose instructions (S1MD instructions), field programmable logic arrays or any mix thereof which provides the desired level performance and cost.

[0123] Aspects of the systems and methods described herein may be implemented as functionalityprogrammed into any of a variety of circuitry, including programmable logic devices (PLDs). such as field programmable gate arrays (FPGAs), programmable array logic (PAL) devices, electrically programmable logic and memory devices and standard cell-based devices, as well as application specific integrated circuits. Some other possibilities for implementing aspects include: memory devices, microcontrollers with memory (e.g., such as EEPROMs, etc.), embedded microprocessors, firmware, software, etc. Furthermore, aspects may be embodied in microprocessors having software-based circuit emulation, discrete logic (sequential and combinatorial), custom devices, fuzzy logic, neural networks, other AI (Artificial Intelligence) or machine learning (ML) systems, quantum devices, and hybrids of any of the above device types.

[0124] It should also be noted that various logic and / or features disclosed herein may be enabled using any number of combinations of hardware, firmware, and / or as data and / or instructions embodied in various machine- readable or computer-readable media, in terms of their behavioral, register transfer, logic component, and or other characteristics. Computer-readable media in which such formatted data and / or instructions may be embodied include, but are not limited to, non-transitory and / or non-volatile media including storage media in tangible various forms (e.g., optical, magnetic or semiconductor storage media, etc.), though do not encompass transitory media.

[0125] Unless the context clearly requires otherwise, throughout the description, the words "comprise," "comprising," and the like are to be construed in an inclusive sense as opposed to an exclusive or exhaustive sense; that is to say, in a sense of "including, but not limited to." Further, the phrase "based on" does not mean "based only on," unless expressly specified otherwise. It will be appreciated by those skilled in the art that various modifications and changes may be made without departing from the scope of the described technology. Such modifications and changes are intended to fall within the scope of the embodiments, it will also be appreciated by those of skill in the art that parts included in one embodiment are interchangeable with other embodiments; one or more parts from a depicted embodiment can be included with other depicted embodiments in any combination. For example, any of the various components described herein and / or depicted in the figures may be combined, interchanged or excluded from other embodiments.

[0126] Words using the singular or plural number also include the plural or singular number respectively. Additionally, the words "herein," "hereunder," "above," "below,” and words of similar import refer to this application as a whole and not to any particular portions of this application. When the word "or" is used in reference to a list of two or more items, that word covers all of the following interpretations of the word: any of the items in the list, all of the items in the list and any combination of the items in the list.

[0127] Other implementations of the inventions will be apparent to those skilled in the art from consideration of the specification and practice of the innovations disclosed herein. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the inventions being indicated by the present disclosure and claims and various associated principles of relevant patent doctrine.

Claims

Claims:

1. One or more non-transitory computer readable media having computer readable instructions stored thereon for realizing a universal (i.e. capable of all neuronal functions possible) software-defined neuron, such as a universal software-defined neuron capable of all neuronal functions possible, wherein the computer readable instructions, upon execution by at least one processor, cause the at least one processor to:implementing the software-defined neuron via executing software modules that create a virtualized central connection control circuit, virtualized registers, and a connection table;transform one or more input signals (I₁.. Iₙ), received at inputs of the software-defined neuron, into one or more output signals (O₁.. Oₙ) to output from the software-defined neuron via a transformation process including:utilizing a logical abstraction layer to process the one or more input signals; processing input parameters of the one or more input signals, where, e.g., the input parameters are represented by real numbers, and transforming the input parameters into input data represented by input floating-point numbers:storing first values of the input floating point numbers in input memory registers; performing, within the logical abstraction layer, a transformation on the input floating-point numbers to produce output floating point numbers, the transformation f representing operation / behavior of the software-defined neuron;storing second values of the output floating point numbers in output memory registers; andgenerating the one or more output signals containing output parameters based on the output floating point numbers, wherein, optionally, the output: parameters are represented by real numbers.

2. The non-transitory computer readable media of claim 1 or the invention of any claim herein, wherein the one or more output signals are configured such that values from the output memory registers are transferred into corresponding ones of the input memory registers when a plurality of software-defined neurons is coupled together.3, The non-transitory computer readable media of claim 1 or the invention of any claim, herein, wherein one or both of the input memory registers and / or the output memory registers provide tuning factors, which either amplify / increase the parameters or cushion / decrease the parameters;wherein, optionally, (i) the tuning point factors include floating-point parameters.4, The non-transitory computer readable media of claim 3 or the invention of any claim herein, wherein one or more of;(i) the tuning factors function as awareness triggers of the software-defined neuron regarding the input signals.(ii) the tuning factors provide relevance triggers of the software-defined neuron to the output signals, and'or, optionally(lit) the tuning factors are configured to provide the software-defined neuron ability to be adapted to learning according to certain tasks during operation, such as via at least one global or local cost function(s) and / or feedback ioop(s);wherein, optionally, the at least one global or local cost function(s) and or feedback loop(s) include, though are not limited to gradient decent algorithms and / or other machine learning procedures implemented between the output and input neurons of a neural or sub neural network. These methods of unconstrained mathematical optimization provide the respective changes for the tuning factors.

5. The lion-transitory computer readable media of claim 1 or the invention of any ciaim herein, wherein one or more of:the tuning factors are utilized to implement adaptive short-term memory functions according to one or more current input signals to make the functions of the whole neural network receptible (receptive?), more or less, to certain sensory or context-based input via local strengthening of connections, e.g„ such as via increasing tuning factors, etc., between neurons that are more frequently used than other connectionsthe tuning factors are configured to be s pread through a whole network of neurons serves as a form of short-term memory of the neural network; and / orchanges to the tuning factors is controlled by centralized functions of the awareness triggers and / or the relevance triggers.

6. The non-transitory computer readable media of claim 4 or the invention of any claim herein, wherein the awareness triggers and / or the rele vance triggers are configured one or both of:as inherent threshold functions of the software-defined neuron: and / orfor utilization to flexibly define the thresholds for the nonlinear activation functions of the software-defined neurons.

7. The non- transitory computer readable media of claim 6 or the invention of any claim herein, wherein the awareness triggers and / or the relevance triggers provide (i) higher tuning factors for more frequently andor extensively used inputs (I) or outputs (O), and / or (ii) lower tuning factors for less used inputs and'or outputs, and, optionally, thereby amplify specific parameters of a central function 310 of a specific software-defined neuron within a software defined neural network (SDNN) during short-term / situation learning.

8. The non-transitory computer readable media of claim 1 or the invention of any claim herein, wherein one or more of:the central connection control circuit of the software-defined neuron utilize a global ID 370 during neural processing;the global ID within the software-defined brain functions as a unique identifier for the software-defined neuron;the connection control circuit pins the global ID to each awareness register and / or to each rele v a ne e reg ister;the global ID is registered within the connection table;the global ID is configured to be registered within connection tables associated with other software-defined neurons:as a function of the global ID being associated with connection tables associated with other software -defined neurons, the software-defined neuron is configured to connect its outputs (O) w ith certain inputs (1) of the other software-defined neurons having different global IDs;as a function of storage and processing of the different global IDs in association with the connection tables of various software-defined neurons, software-defined connected outputs of software-defined neurons pass their parameter values on to inputs of other software-defined neurons;a connection between two specific software-defined neurons is established by a pair of global ID entries within the connection tables of the two specific software-defined neurons, each respectively having each other's global IDs stored; In this manner, the connection matrix 250 is established.a connection matrix is established based on sets of connections between two specific software-defined neurons established by a pair of global ID entries within the connection tables of the two, specific software-defined neurons, each respectively having each other’s global IDs stored;the transformation process and / or aspects related to the connection matrix are characterized via information regarding software-defined neuron registers set forth in Table 1.

9. The non-transitory computer readable media of claim 1 or the invention of any claim herein, wherein the transformation process further includes:implementing a global ID 370 within the software-defined brain that functions as a unique identifier for the software-defined neuron; and / orutilizing, by the central connection control circuit, the global ID to pin the global ID to at least one, or to each, awareness register and / or to at least one, or to each, relevance register.

10. The non-transitory computer readable media of claim 1 or the invention of any claim herein, wherein one or more of:the steps for implementing the software-defined neuron are configured for execution via a hardware implementation that supports an algorithm for the transformation process;one or both of the input parameters and / or the output parameters represent numbers that are complex though are each compressible into a single real number, thereby enabling representation in the software-defined neuron by the floating-point numbers within the logical abstraction layer;the transformation process is configured to process the input signals comprised of physical signals containing physical representations of data indicative of physical properties:the output signals provided for transfer between tire software-defined neurons are configured as a function of a specific hardware implementation and / or related overall performance characteristics, though have no impact to the function of the software-defined brain;the physical signals are configured to convey data via physical properties, such as one or more of time coding, phase shift, amplitude modulation, frequency modulation, quantum entanglement, and / or quantum superposition, among other physical properties; and-'orthe transformation process is configured to process physical features (e.g., the physical signals, the physical representations, the physical properties, the data indicative of physical properties, etc,) of the input signals to derive measurements of real numbers which are to be converted into the floating-point numbers for the software-defined neurons.

11. The non-transitory computer readable media of claim I or the invention of any claim herein, wherein the transformation process includes:trans forming the input signals into the output signals that comprise output parameters Oxhaving configuration such that each output signal is configured: (i) to have connection to exactly only one of the input signals of another one of a plurality of software-defined neurons, and (ii) to make the another one identical to the software-defied neuron via the connection and the configuration.

12. One or more non-transitory computer readable media having computer readable instructions stored thereon for implementing a software-defined neural network, wherein the computer readable instructions, upon execution by at least one processor, cause the at least one processor to:implement a plurality of software-defined neurons in an array forming an architecture of the software defined neural network, each of the software-defined neurons configured to transform one or more input signals (I1.. In) into one or more output signals (O1.. Ox) via a transformation process, which may optionally include one or more of:utilizing a logical abstraction layer to process the one or more input signals; processing input parameters of the one or more input signals, where, e.g,. the input parameters are represented by real numbers, and transforming the input parameters into input data represented by input floating-point numbers;storing first values of the input floating point numbers in input memory registers;performing, within the logical abstraction layer, a transformation fion the input floating-point numbers to produce output floating point numbers, the transformation representing operation, 'behavior of the software-defined neuron;storing second values of the output floating point numbers in output memory registers; and / orgenerating the one or more output signals containing output parameters based on the output floating point numbers, wherein, optionally, the output parameters are represented by real numbers;couple outputs of the plurality of software-defined neurons with inputs of other software- defined neurons in the array;generate, in connection with the output signals from the output memory registers, output data with the outputs, the output data including the output floating point numbers being stored in the output memory registers, the output floating numbers being provided via the outputs defined as parameters of the outputs;provide each one of the outputs including the output floating numbers and the parameters as an input to an adjacently connected software-defined neuron; andimmediately set up the parameters of each said one of the outputs as an input value of the adjacently connected software-defined neuron.

13. The non-transitory computer readable media of claim 12 or the invention of any claim herein, wherein the computer readable instructions further include instructions that, upon execution by at least one processor, cause the at least one processor to:provide control instructions to a matrix of the software-defined neurons defined as having a first dimension (e.g. width) defined as ‘n’ lines and a second dimension (e.g. depth) defined as ‘in’ rows of the software-defined neurons; andestablish connections between internal cells of the ‘o ' lines with the ‘m’ rows of other ones of the internal cells of the matrix of the software-defined neurons, to form an n by m (N x M) nexus matrix; andconfigure free inputs Ipand free outputs Oqas interfaces for connection to other software- defined neural networks or other systems outside of the matrix of the software-defined neurons.

14. The non-transitory computer readable media of claim 12 or the invention of any claim herein, wherein, machine conscious and subconscious functions of arbitrary heterogeneous compositions of different types of neural networks based on the uni versal substrate of the software-defined neural network, such as that of claims 1-11, are introduced with the hierarchical order of information processing, e.g., optionally the heterogeneous hierarchical model (HHM) as the universal description of the software-defined brain, wherein the software-defined neural networks are configured to provide various different types of sensory perception (such as hearing, seeing, feeling,...), pattern recognition(such as recognizing known patterns (such as sounds, forms, touches,...), memory functions (e.g. to associate previous experiences or feelings), inherent reactions (such as reflexes or instincts), subconscious reasoning to make sense of the mixture of sensory input and associated pre-known information, together with conscious generative composition of the outside world to predict future events therein and compare these predictions within comparison networks with actual sensory input, the generation of language composition, language decomposition, short-term memory, attention, and other cognitive functions such as physical control of machines or bodies by any nature.

15. The non-transitory computer readable media of claim 12 or the invention of any claim herein, wherein one or more of:within the software-defined neural network, the software-defined neurons are specifically connected to each other at any instance of time;other than in a situation learning-short-term memory, the connections between the software- defined neurons are not subject to change during the process of inference of the software-defined neural network:

16. The non-transitory computer readable media of claim 12 or the invention of any claim herein: wherein, functions associated with the software-defined neurons may become embedded matrix representations (MRs), e.g., optionally associated with the connection matrices, in the software- defined neural networks; and / orwherein, the matrix representations include sets of embedded vectors which in turn represent data types for a given function to be applied on input data in the software-defined neural network.

17. One or more non-transitory computer readable media having computer readable instructions stored thereon for implementing a software-defined brain, wherein the computer readable instructions, upon execution by al least one processor, cause the at least one processor to:implement an array of software-defined neural networks forming an architecture of lhe software-defined brain, each of the software-defined neural networks comprised of an array of a plurality of software-defined neurons;configuring a plurality of software-defined neural networks that are differently specialized in that each specialized neural network is configured with one or more of different connectivity, different behavior in time, and / or different input and / or output characteristics; coupling together the plurality of software-defined neural networks that are differently specialized:partitioning at least a first subset of the software-defined neural networks to execute conscious functions of the software-defined neural networks;partitioning at least a second subset of the software-defined neural networks to execute subconscious functions of the software -defined neural networks; andimplementing the software-defined brain to execute the conscious functions of the first subset as well as the subconscious functions of the second subset;wherein, optionally, the software-defined brain is implemented as a function of separation configuration principles that (i) each sub neural network exclusively provides functionality either to the conscious or alternatively to the subconscious part of the software- defined brain, and (ii) each function of consciousness or subconsciousness is configured to be separately and subsequently implemented into a software-defined brain.

18. The non-transitory computer readable media of claim 17 or the invention of any claim herein, wherein one or more of:configuration of the specialized neural networks with different connectivity includes implementation of:hierarchical neural networks (or, e.g., hierarchically connected neurons, neurons never connected to a previous layer, etc.), andconvolutional neural networks (or, e.g., convolutionally connected neurons, neurons occasionally connected to a previous layer, etc.};configuration of the specialized neural networks with different behavior in time includes implementation of two or more of:synchronous neural networks (or, e.g., synchronized by a standard clock, output parameters of one layer of the network are fully computed and transferred to all input lines of the next layer before the calculation starts, etc,);asynchronous neural networks (or, e.g., absence of a synchronizing clock, different calculation times of the neuronal functions, different time delay of signals via the connection lengths between the outputs and inputs of the neurons, etc.); and / or chaotic neural networks (or, e.g., the hardware does not even provide deterministic calculation times for the neuron functions, no stable signal transfer speed between the neurons exists, etc.); and / orconfiguration of the specialized neural networks with different types of input characteristics and output characteristics includes implementation of two or more of:linear sequential networks or neural networks that use sequential input to compute output as pieces of information, called token;linear tensor networks or providing whole tokens or tensors of information at once as a whole to input neurons;recursive sequential networks (e.g., classical neural networks, such as GPTs, networks which compute sequential input and output of pieces of information (tokens), networks that feed parts or all of their output tokens back into their input neurons for the next round of computation, etc,);recursive tensor networks or networks that feed their outputs partly or totally backio their inputs to create a constant loop of adaption;wherein, as a function of said configuration of the specialized neural networks, the software- defined brain constructed from the plurality of software-defined neural networks is heterogenous as a function of one or more of connectivity, behavior in time, and / or different types of input characteristics and output characteristics, referred to herein as a ‘heterogeneous hierarchical model. ’19. The non-transitory computer readable media of claim 17 or the invention of any claim herein, wherein one or both of:within the software-defined neural network, the software-defined neurons have specific connections to each other at any instance of time;other than in instance(s) of situation learning / short-term memory, the specific connections between the software-defined neurons are not subject to change during the process of inference of the software-defined neural network.

20. The non-transitory computer readable media of claim 17 or die invention of any claim herein:wherein the conscious functions include one or more of personality, desire, intent, contextualization, short-term memory, attention, free will, feelings, and / or event model projection, etc.; and / orwherein the subconscious functions include one or more of perception, conception, reasoning, intermediate memory, long-term memory, memory management, and / or positioning, etc.;wherein, optionally, the conscious functions may be implemented via a direct actuator network 460 connected to positioning in order to minimize the time for the free will to take physical action.

21. The non-transitory computer readable media of claim 17 or the invention of any claim herein, wherein the software-defined brain is further implemented to comprise one or more transcendent functional groups such as though not limited to multiple personalities, common perception, memory mapping, hierarchical event model projection, mind cloning, and / or mind merging.

22. One or more non-transitory computer readable media having computer readable instructions stored thereon for implementing functions or workflows for machine consciousness associated with a software-defined brain and. or a heterogeneous hierarchical model (HHM), wherein the computer readable instructions, upon execution by at least one processor, cause the at least one processor to:implement an array of software-defined neural networks forming an architecture of the software-defined brain, each of the software-defined neural networks comprised of an array of a plurality of software-defined neurons;configuring a plurality of software-defined neural networks that are differentlyspecialized in that each specialized neural network is configured with one or more of different connectivity, different behavior in time, and / or different input and / or output characteristics; configuring one or more of the pluralities of software-defined neural networks with computer program instructions that perform workflow for data and functions and. upon executing by the at least one processor, implement one or more of a data encoder networks 3110, a function encoder network 3130, and / or a convergent decoder network 3140 to provide at least one software-defined functional neural network that generates meta data outputs based on data input thereto;coupling together the plurality of software-defined neural networks that are differently specialized;performing processing regarding execution of conscious functions and subconscious functions of the software-defined brain to provide output data based on processing of input data via the data encoder network 3110, the function encoder network 3130, and / or the convergent decoder network of the at least one software-defined functional neural network.

23. The non-transitory computer readable media of claim 22 or the invention of any claim herein, wherein one or more of:the data encoder network 3110 is configured to embed one or more vector representations of the input data 3120;the function encoder network 3130 is configured to embeds one or more matrix representations of at least one software-defined brain or neural network function:the convergent decoder network 3140 is configured to realize computation of the at least one software-defined brain or neural network function on the data 3150.

24. The non-transitory computer readable media of claim 22 or the invention of any claim herein, wherein, as a function of configuring the software-defined brain via a plurality of specialized neural networks, the software-defined brain constructed from the plurality of software-defined neural networks is heterogenous as a function of one or more of connectivity, behavior in time, and'or different types of input characteristics and output characteristics, referred to herein as a heterogeneous hierarchical model (HHM).

25. One or more non-transitory computer readable media having computer readable instructions stored thereon for implementing machine consciousness workflows associated with a software-defined brain, wherein the computer readable instructions, upon execution by at least one processor, cause the at least one processor to:implement an array of software -defined neural networks forming an architecture of the software-defined brain, each of the software-defined neural networks comprised of an array of a plurality of software-defined neurons;con figuring a plurality of software-defined neural networks that are differently specialized in that each specialized neural network is configured with one or more of different connectivity, different behavior in time, and / or different input and / or output characteristics: coupling together the plurality of software-defined neural networks that are differently specialized; andperforming processing regarding execution of machine conscious functions and machine subconscious functions of the software-defined brain to provide output data based on processing of input data via the plurality of differently specialized software-defined neural networks: wherein machine conscious functions are configured as recursive 450,460 and generative composition (cognitive function) 481, 482 based on sensation and memory 441, 442, 443; and / orwherein underlying cognitive functions 470 are hidden to the machine conscious functions thereby making the underly ing cognitive functions become machine subconscious functions.

26. The non-transitory computer readable media of claim 25 or the invention of any claim herein, wherein the subconscious functions of the software-define neural networks comprise local computational tasks, which run in parallel and are synchronized with each other, optionally, (i) such that one single instance of a global computational task is / becomes able to predict what will happen during a next moment within its worldview, and / or ( ii) to constantly compare its assumptions with the incoming data to adapt its digital twin of its own body and the outside world according to reality.

27. The non-transitory computer readable media of claim 25 or the invention of any claim herein, wherein;performing continuous processing of data aggregation and analysis on many local (neural network) levels, which ultimately leads to information processing on a global level of the software-defined brain where a constant alignment between short-term prediction by generative composition 481 and the varying sensing data happens within the actuator 460 and comparison neural networks 450.

28. The non-transitory computer readable media of claim 27 or the invention of any claim herein, wherein:wherein, during said continuous processing, the generative composition 481 performs neural processing to predict a near future result of an outside environment by construction of a digital twin (a virtual model) of the outside environment which is matched with sensory data and accordingly adjusted so that the difference function is minimized.

29. The non-transitory computer readable media of claim 27 or the invention of any claim herein, wherein the performing continuous processing includes:performing, via the comparison neural network 450, neural processing to predict a near future result of an outside environment by construction of a digital twin (a virtual model) of the outside environment which is matched with sensory data and accordingly adjusted so that the difference function is minimized; andreproducing, via the generative composition 481, a constantly updated model of the outside environment in order to make the inherent processing (machine mind) of the software-defined brain to orient oneself.

30. The non-transitory computer readable media of claim 25 or the invention of any claim herein, wherein the plurality of software-defined neural networks includes one or more first groups of neural networks for workflow of machine-subconsci ous neural networks and one or more second groups of neural networks for workflow of machine-conscious neural networks.

31. The non-transitory computer readable media of claim 30 or the invention of any claim herein, wherein the one or more first groups of neural networks comprise locally represented machine subconscious neural networks, and wherein, optionally, the one or more first groups comprise four groups of machine subconscious neural networks including:perception neural networks, such as those associated with the data analyzer 430 and / or that provide data clearance and reduction so that subsequent, higher order brain functions can handle the information flow;conception neural networks, such as those associated with pattern recognition 441, inherent algorithms 442, and / or long-term memory 443, which, optionally, include neural networks that one or more of (i) identify known input patterns, (ii) warn if there is no match with anything known, and / or then (iii) draw attention of higher-level software-defined brain functions to the fact via the actuator network 460:reasoning neural networks, such as those associated with the reasoning neural networks 470, which, optionally, include neural networks that provide quick, machine-subconscious assessments and classify the situations but tend to bias the machine-conscious mind with stereotyped thinking in more complex cases; and / orpositioning neural networks, such as those associated with the physical control neural networks 490 and may be activated by the actuator network 460,32. The non-transitory computer readable media of claim 30 or the invention of any claim herein, wherein the one or more second groups of conscious neural networks compri se globally represented conscious neural networks, and wherein, optionally, the one or more second groups comprise one group of conscious neural networks including:comprehension neural networks, such, as those associated with generative composition 481, which may, optionally, include neural networks that establish a digital twin simulation about the outside environment or retrieve said digital twin from short-term or long-term memory, which may then be aligned with comparison 450 neural networks which provide a divergence function between the sensor systems and the short-term prediction of the simulation which then is subject to minimization by the regulative actuator network 460 in accordance with the generative composition.

33. The non-transitory computer readable media of claim 25 or the invention of any claim herein, wherein the software-defined brain is implemented utilizing physical control neural networks, wherein, optionally, in new, unlearned situations, the physical control neural networks rely on detailed actuator information control by the generative composition 481.

34. The non-transitory computer readable media of claim 33 or the invention of any claim herein, wherein the software-defined brain is implemented such that, for often repeated physical control or inherent algorithms 442, the machine-subconscious actuator signals are sufficient for the physical control neural network, which, optionally, then enriches stimulus with (necessary, detailed) formerly trained information, and thereby, in some embodiments of such cases, the physical control network does not need machine-conscious attention.

35. One or more non-transitory computer readable media having computer readable instructions stored thereon for implementing machine workflows associated with a software-defined brain, wherein the computer readable instructions, upon execution by at least one processor, cause the at least one processor to:implement an array of software-defined neural networks forming an architecture of the software-defined brain, each of the software-defined neural networks comprised of an array of a plurality of software-defined neurons;configure a plurality of the software-defined neural networks that are differently specialized in that each specialized neural network is configured with one or more of different connectivity, different behavior in time, and / or different input and / or output characteristics: couple together the plurality of software-defined neural networks that are differently specialized;assign a first set of the software-defined neural networks to perform machine conscious functions and a second set of the software-defined neural networks to perform machine subconscious functions:train the plurality of software-defined neural networks and / or the software-defined brain by:providing prepared training data as an input matrix to input interfaces of neuralnetworks to be trained;analyzing the prepared training data via a subject neural network being trained as it ingests the prepared training data during training;storing initial results data generated via output interfaces of the neural networks being trained within one or more output matrices;transferring the initial results data from the one or more output matrices to a rating system;comparing, via the rating system, the initial results data against desired values of the subject neural networks’ desired processing associated wi th the software- defined brain;performing an incremental training process of the subject neural network until final training results of the subject neural network correspond to the desired values for the subject neural network; andimplementing the subject neural network that’s been trained as a software- defined neural network of the software-defined brain.

36. The non-transitory computer readable media of claim 35 or the invention of any claim herein, wherein the performing an incremental training process of the subject neural network includes:based on results of the comparing, providing feedback information, one or both of directly to the subject neural network being trained or indirectly via the prepared training data, to incrementally train each the subject neural network with the feedback information;storing incremental results of the training within the one or more output matrices; processing, via the rating system, the incremental results against the desired values: and repeating the incremental training process until the incremental results correspond to the desired values for the subject neural network.

37. The non- trans itory computer readable media of claim 35 or the invention of any claim herein, wherein feedback being provided to the subject neural network via the training is provided as direct feedback that is provided directly to the subject neural network;wherein, optionally, may be configured to influence one or more of tuning factors 340, awareness 350 and relevance triggers 360 of the subject neural network and / or connections 390 between software-defined neurons of the subject neural network.38, The non- trans itory computer readable media of claim 35 or the invention of any claim herein, wherein feedback being provided to the subject neural network via the training is provided as indirect feedback, such as feedback managed by a data preparation unit 510 that is constructed to generate variations of input matrices configured to produce a match between the incremental results and the desired values via one or more iterations of the incremental training process;wherein, optionally, the data preparation unit is configured to stimulates inherent threshold functions, such as associated with awareness 350 and relevance 360, of the software-defined neurons within the subject neural network, results in a gradual reorganization of the subject neural network during training and / or permanent learning results thereof.

39. One or more non-transitory computer readable media having computer readable instructions stored thereon for implementing a software-defined brain, wherein the computer readable instructions, upon execution by at least one processor, cause the at least one processor to:implement an array of software-defined neural networks forming an architecture of the software-defined brain, each of the software-defined neural networks comprised of an array of a plurality of software-defined neurons;configure a plurality of the software-defined neural networks that are differently specialized in that each specialized neural network is configured with one or more of different connectivity, different behavior in time, and / or different input and / or output characteristics; couple together the plurality of software-defined neural networks that are differently specialized; andone or more of:assign a first set of the software -defined neural networks to perform machine conscious functions and a second set of the software-defined neural networks to perform machine subconscious functions;utilize the plurality of the software-defined neural networks to implement the software- defined brain, including one or more of:receiving input information, e.g., via input matrices, from one or more mechanical body sensory components 620 associated with one or more machine subconscious and / or machine conscious functions that are embedded in and / or associated with a mechanical body or mechanical elements associated with the software-defined brain;analyzing andor interpreting the input information together with internal state information of the software-defined brain;generating, based on the analyzed and / or interpreted information, a constant output that is translated into actions in a physical environment 640 via interfaces 630 that are embedded in and / or associated with the mechanical body or the mechanical elements.

40. The non-transitory computer readable media of claim 38 or the invention of any claim herein, wherein the computer readable instructions include instructions that, upon execution by at least one processor, further cause the at least one processor to:train the plurality of software-defined neural networks and / or the software-defined brain by:providing prepared training data as an input matrix to input interfaces of neural networks to be trained;analyzing the prepared training data via a subject neural network being trained as it ingests the prepared training data during training;storing initial results data generated via output interfaces of the neural networks being trained within one or more output matrices;transferring the initial results data from the one or more output matrices to a rating system;comparing, via the rating system, the initial results data against desired values of the subject neural networks’ desired processing associated with the software- de fined brain;performing an incremental training process of the subject neural network until final training results of the subject neural network correspond to the desired values for the subject neural network; andimplemciiting the subject neural network that’s been trained as a software- defined neural network of the software-defined brain.wherein, optionally, the performing an incremental training process of the subject neural network includes one or more of:based on results of the comparing, providing feedback information, one or both of directly to the subject neural network being trained or indirectly via the prepared training data, to incrementally train each the subject neural network with the feedback information;storing incremental results of the training within the one or more output matrices; processing, via the rating system, the incremental results against the desired values: and / or repeating the incremental training process until the incremental results correspond to the desired values for the subject neural network.

41. One or more non-transitory computer readable media having computer readable instructions stored (hereon for realizing a universal (i.e. capable of all neuronal functions possible) software-defined neuron, wherein the computer readable instructions, upon execution by at least one processor, cause the at least one processor to:implementing the software-defined neuron via executing software modules that create a virtualized central connection control circuit, virtualized registers, and a connection table;transform one or more input signals (I,.. I„), received at inputs of the software-defined neuron, into one or more output signals (O;.. O») to output from the software-defined neuron via a transformation process including:processing input signals including input parameters and / or input data of the input signals;storing the input parameters and / or the input data in input memory registers;performing a transformation f on the input parameters and / or the input data to produce output data, the transformation representing operation-behavior of the software-de fined neuron;storing the output data in output memory registers;generating the one or more output signals containing the output data: and wherein one or both of the input memory registers and / or the output memory registers provide tuning factors, which either amplify / increase the parameters or cushion / decrease the parameters, wherein one or more of: (i) the tuning factors function as awareness triggers of the software-defined neuron regarding the input signals, (ii) the tuning factors provide relevance triggers of the software-defined neuron to the output signals, and / or (iii) the tuning factors are configured to provide the software-defined neuron ability to be adapted to learning according to certain tasks during operation, such as via at least one global or local cost function(s) and / or feedback loop(s);wherein, optionally, the at least one global or local cost function(s) and / or feedback loop(s) include, though are not limited to gradient decent algorithms and / or other machine learning procedures implemented between the output and input neurons of a neural or sub neural network; and'orwherein, via such unconstrained optimization, respective changes for the tuning factors are provided.

42. One or more non-transitory computer readable media having computer readable instructions stored thereon for implementing a software-defined neural network, wherein the computer readable instructions, upon execution by at least: one processor, cause the at least one processor to:implement a plurality of software-defined neurons in an array forming an architecture of the software-defined neural network, each of the software-defined neurons configured to transform one or more input signals (I1.. I») into one or more output signals (Oi.. O») via a iraasformation process including:processing input signals including input parameters and / or input data of the input signals:storing the input parameters and / or the input data in input memory registers; performing a transformation f on the input parameters and / or the input data to produce output data, the transformation f representing operation / behavior of the software-defined neuron;storing the output data in output memory registers; andgenerating the one or more output signals containing the output data:wherein one or both of the input memory registers and / or the output memory registers provide tuning factors, which either amplify / increase the parameters or cushion / decrease the parameters, wherein one or more of: (i) mathematicalprocessing (proof via such processing) tor said universal character about any possible limctionality gained by an arbitrary neural network (SDNN) constructed of such SDNONs tuning factors function as awareness triggers of the software-defined neuron regarding the input signals, (ii) the tuning factors provide relevance triggers of the software-defined neuron to the output signals, and / or ( iii) the tuning factors are configured to provide the software-defined neuron ability to be adapted to learning according to certain tasks during operation, such as via at least one global or local cost function(s) and / or feedback loop(s);wherein, optionally, the at least one global or local cost function(s) and / or feedback loop(s) include, though are not limited to gradient decent algorithms and / or other machine learning procedures implemented between the output and input neurons of a neural or sub neural network;couple outputs of the plurality of software-defined neurons with inputs of other software- defined neurons in the array;wherein universality of the software-defined neural network is achieved via one or more of:partitioning at least a first subset of the software-defined neural networks to execute machine conscious functions of the software-defined neural networks;partitioning at least a second subset of the software-defined neural networks to execute machine subconscious functions of the software -defined neural networks;the software-defined neural networks being implemented as a function of separation configuration principles that (i) each sub neural network exclusively provides functionality either to the machine conscious or alternatively to the machine subconscious part of the software- defined brain, and (ii) each function of machine consciousness or machine subconsciousness is configured to be separately and subsequently implemented into a software-defined brain;the machine conscious functions being implemented via an actuator network 460 having one or more hidden layers, wherein the actuator network is configured to maintain compatible software-defined neurons both on their input and output layers to meet different signal encoding for respective neural networks to connect, wherein the hidden layers are configured to and control processing necessary exact signal conversion(s) therebetween, which cannot be done by any other neural network: with different: functions.: and / orvia operation of the actuator network, the actuator network generates / renders output parameters that are compatible with input parameters of a following network, such as via immediately setting up the parameters of each said one of the outputs as an input value of the adjacently connected software-defined neuron(s).

43. One or more non-transitory computer readable media having computer readable instructions stored thereon for implementing a software-defined brain, such as via utilization of a heterogeneous hierarchical model (HHM). wherein the computer readable instructions, upon execution by at least oneprocessor, cause the at least one processor to:implement an array of software-defined neural networks forming an architecture of the software-defined brain, each of the software-defined neural networks comprised of an array of a plurali ty of software-defined neurons;configure a plurality of software-defined neural networks that are differently specialized in that each specialized neural network is configured with one or more of different connectivity, different behavior in time, and / or different input and / or output characteristics; andone or more of:partition at least a first subset of the software-defined neural networks to execute conscious functions of the software-defined neural networks;partition at least a second subset of the software-defined neural networks to execute subconscious functions of the software-defined neural networks; andperform processing regarding execution of conscious functions and subconscious functions of the software-defined brain to provide output data based on processing of input, data via a data encoder network 3110, a function encoder network 3130, and / or a convergent decoder network associated with one or more of the software-defined neural networks; and / or couple together the plurality of software-defined neural networks that are differently specialized to create the heterogeneous hierarchical model (HHM as the universal description of the software-defined brain, wherein the software-defined neural networks are configured to provide various different types of sensory perception (e.g., hearing, seeing, feeling, etc.), pattern recognition (e.g,, recognizing known patterns, sounds, forms, (ouches, etc,), memory functions (e.g. that associate previous experiences or feelings, etc.), inherent reactions (e.g., reflexes, instincts, etc,), subconscious reasoning to make sense of the mixture of sensory input and associated pre-known information, together with conscious generative composition of the outside world to predict future events therein and compare these predictions within comparison networks with actual sensory input, the generation of language composition, language decomposition, short-term memory, attention. and other cognitive functions such as physical control of machines or bodies by any nature.44, One or more non-transitory computer readable media having computer readable instructions stored thereon for implementing a software-defined functional neural network associated with a software- defined brain, wherein the computer readable instructions, upon execution by at least one processor, cause the al least one processor to:implement an array of software-defined neural networks forming an architecture of the software-defined brain, each of the software-defined neural networks comprised of an array of a plurality of software-defined neurons;wherein functions associated with the software-defined neurons are configured as embedded matrix representations (MRs) such as in connection matrices associatedwith the software-defined neural networks:wherein the matrix representations include sets of embedded vectors which in turn represent data types for a given function to be applied on input data in the software-defined neural network;configuring a plurality of software-defined neural networks that are differently specialized in that each speci alized neural network is configured wi th one or more of di fferent connectivity, different behavior in time, and / or different input and / or output characteri tics: configuring one or more of the pluralities of software-defined neural networks with computer program instructions that perform workflow for data and functions and, upon executing by the at least one processor, implement one or more of a data encoder networks 3110, a function encoder network 3130, and / or a convergent decoder network 3140 to provide one or more software-defined functional neural networks that generate meta data outputs based on data input thereto;coupling together the plurality of software-defined neural networks that are differently specialized;performing processing regarding execution of machine conscious functions and machine subconscious functions of the software-defined brain to provide output data based on processing of input data via a function encoder network 3130 that, is configured to embed the matrix representations associated with the functions performed on the input data via the software- defined brain.

45. One or more non-transitory computer readable media having computer readable instructions stored thereon for implementing a software-defined brain via a heterogeneous hierarchical model (HHM). wherein the computer readable instructions, upon execution by at least one processor, cause the at least one processor to:implement an array of software-defined neural networks forming an architecture of the software-defined brain, each of the software-defined neural networks comprised of an array of a plurality of software-defined neurons;configure a plurality of software-defined neural networks that are differently specialized in that each specialized neural network is configured with one or more of different: connectivity, different behavior in time, and / or different input and / or output characteristics;partition at least a first subset of the software-defined neural networks to execute machine conscious functions of the software-defined neural networks:partition at least a second subset of the software-defined neural networks to execute machine subconscious functions of the software-defined neural networks: andperform processing regarding execution of machine conscious functions and machine subconscious functions of the software-defined brain to provide output data based on processing of input data of the software-defined neural networks;wherein the software-defined neural networks are implemented as a function of separation configuration principles that (i) each sub neural network exclusively provides functionality either to the machine conscious or alternatively to the machine subconscious part of the software-defined brain, and (ii) each function of machine consciousness or machine subconsciousness is configured to be separately and subsequently implemented into a software- defined brain;wherein the machine conscious functions are implemented via an actuator network 460 having one or more hidden layers, wherein the actuator network is configured to maintain compatible software-defined neurons both on their input and output layers to meet different signal encoding for respective neural networks to connect, wherein the hidden layers are configured to and control processing necessary for exact signal conversion(s) therebetween.

46. One or more non -transitory computer readable media having computer readable instructions stored thereon for implementing a software-defined brain and-'or a software-defined actuator neural network, wherein the computer readable instructions, upon execution by at least one processor, cause the at least one processor to:implement an array of software-defined neural networks forming an architecture of the software-defined brain, each of the software-defined neural networks comprised of an array of a plurality of software-defined neurons;configure a plurality of the software-defined neural networks that are differently specialized in that each specialized neural network is configured with one or more of different connectivity, different behavior in time, and / or different input and'or output characteristics; couple together the plurality of software-defined neural networks that are differently specialized; andone or more of:assign a first set of the software-defined neural networks to perform machine conscious functions and a second set of the software-defined neural networks to perform machine subconscious functions:implementing the actuator neural network via one or more of:configuring the actuator neural network consistent with one or more of Figure 4A, Figure 4B, Figure 3 A, Table 1, Table 2, the above detailed description associated with same, and / or with other disclosure of the written description or claims herein;the machine conscious functions being implemented via the actuator neural network having one or more hidden layers, and / or wherein the actuator network is configured to maintain compatible software-defined neurons both on their input and output layers to meet different signal encoding for respecti ve neural networks to connect, and / or wherein the hidden layers are configured to and control processing necessary exact signal conversion(s) therebetween, which cannot be done by any other neural network with different functions; and / oropera tion of the actuator neural network, wherein the actuator neural network generates / renders output parameters that are compatible with input parameters of a following network, such as via immediately setting up the parameters of each said one of the outputs as an input value of the adjacently connected software-defined neuron(s); and / or utilize the plurality of the software-defined neural networks to implement the software- defined brain, including one or more of:receiving input information, e.g.. via input matrices, from one or more mechanical body sensory components 620 associated with one or more machine subconscious and / or machine conscious functions that are embedded in and / or associated with a mechanical body or mechanical elements associated with the software-defined brain:analyzing and / or interpreting the input information together with internal state information of the software-defined brain;generating, based on the analyzed and / or interpreted information, a constant output that is translated into actions in a physical environment 640 via interfaces 630 that are embedded in and / or associated with the mechanical body or the mechanical elements.

47. A system comprising:at least one processor; andone or more non-transitory computer readable media having computer readable instructions stored thereon, wherein the computer readable instructions, upon execution by the at least one processor, cause the at least one processor to:implement one or more operations of the computer readable media recited in any of claim 1 through claim 46.

48. A system comprising:at least one processor; andone or more non-transitory computer readable media having computer readable instructions stored thereon, wherein the computer readable instructions, upon execution by the at least one processor, cause the at least one processor to:perform one or more operations and / or steps associated with and or involving one or more features, portions, processes, systems, methods and'or aspects set forth in any claim and / or portion of claim herein and / or implemented via one or more figures, tables, elements, componen ts, subcomponents, aspects, steps, features and / or functionality of the disclosed technology set forth elsewhere herein.

49. A method comprising:implementing one or more operations, and'or part or portion thereof, of the computer readable media recited in any of claim 1 through claim 46.

50. A method comprising:performing one or more steps, processes, operations, features and / or functionality associated with and / or involving one or more computer-readable media process, feature, portion, operation, system, method and / or aspects set forth in any claim and / or in any portion of claim herein and / or implemented in one or more figures, tables, elements, components, subcomponents, aspects, steps, features and / or functionality of the disclosed technology set forth elsewhere herein.