Cognitive computing methods and systems based on biological neural networks

Abstract

A biological neural network (BNN) core unit, which includes a neural cell culture, an input stimulation unit, and an output readout unit, can be controlled through its various life cycles to provide data processing functions. An automated system, which includes an environmental and chemical controller unit suitable for operating with the BNN stimulation and readout data interface, facilitates the monitoring and adjustment of the BNN core unit parameters. Pre-processing and post-processing of the BNN interface signals can further facilitate the training and reinforcement learning of the BNN. Multiple BNN core units can also be assembled together as a stack. The proposed system provides a BNN operating system as a core component for a wetware server to receive, process and send data for different client applications without exposing the BNN core unit components to client users, while requiring significantly less energy than traditional silicon-based hardware and software information processing for advanced cognitive computing tasks.

Description

Technical Field

[0001] This disclosure relates to cognitive computing systems, methods, and processes that perform a variety of advanced, complex cognitive tasks that mimic and extend biological brain functions. The proposed cognitive computing systems, methods, and processes utilize multiple neural cells as their biological core processing elements combined with brain-on-a-chip interfaces and controllers to complement and interact with more traditional information technology networking and computing architectures. Background Art

[0002] With the tremendous advancements in information technology (IT) over the past few decades, numerous methods and systems are now available for performing a wide range of computing tasks, such as computation, data optimization, data classification, natural language processing and translation, and image and video processing and recognition. Recent developments in cognitive computing, including machine learning, deep learning, and artificial neural computing (ANC), which mimics biological neural networks, have further advanced the field. ANC further seeks to provide humans with artificial intelligence to assist them in tasks such as object and facial recognition, natural language processing (NLP) and sentiment analysis, and in hostile environments such as the deep sea and space, as well as in nuclear reactors. Consequently, major companies such as Google (Google Cloud Platform), Amazon (Amazon Web Services AWS), and Microsoft (Azure) offer high-performance computing, high-throughput computing, and high-availability systems and infrastructure as cloud computing services. Facebook and Apple also operate their own private data centers at carefully selected data farms worldwide, where they benefit from affordable and reliable electricity. However, a major limitation of current silicon-based computing environments (both software and hardware) is the excessively high power required to perform complex cognitive tasks. Silicon-based systems consume an order of magnitude more power than biological systems with similar performance, such as the brain. Another limitation is the requirement for at least some explicit logic programming and structured data representation, which makes them unsuitable today for implementing concepts that are not yet understood and modeled by human science, such as higher-level cognitive processes, creative thinking, and consciousness.

[0003] Neural networks can be viewed as a means of creating a spatiotemporal mapping between two spaces of different dimensions. In configurations where the number of outputs is smaller than the number of inputs, neural networks can thus create simplified representations of problems in a space with finite dimensions. For example, well-known spatial transforms include the Hough transform, which maps a two-dimensional space to a two-parameter voting space; the Fourier transform, which maps a time-periodic signal to its frequency representation; and wavelet transforms, which map images to a scale-space representation. While these transforms are well-described mathematically and straightforward to implement, even if neural networks could be used, they are unlikely to be the most computationally efficient choice. However, in cases where, for example, a 1000-dimensional multivariate input space produces only 10 correlated variables at output, an explicit mathematical solution is often too complex. Neural networks offer a promising approach to solving these problems. However, the internal state space dimensions required to solve such problems can be enormous and require excessive computational resources on current hardware.

[0004] As an alternative to traditional software and hardware information technology, as early as the end of the 20th century, scholars mainly explored wet software solutions based on biological components (such as cultured cells) instead of transistors ( https: / / www.technologyreview.com / s / 400707 / biologicalcomputing / Until now, most scientists have focused on how to achieve basic functions similar to basic core IT processing logic gates, calculations and memory storage, such as DNA computing ( https: / / www.nature.com / subjects / dna-computing While this path holds great promise, especially with recent advances in synthetic biology and DNA editing, engineering higher-order cognitive processes from these fundamental functions remains as challenging as with traditional silicon-based logic.

[0005] An example of advanced cognitive capabilities is the functionality required for a generally intelligent machine to pass the Turing test. Current research using artificial neural networks (ANNs) in this regard faces at least two major limitations:

[0006] Compared to biological neural networks (BNNs), existing multi-layer networks used for deep learning lack temporal flexibility. The human brain is essentially a biological multi-core system, for which no mathematical model exists. Most of the powerful mathematics used in the greatest scientific models, such as quantum mechanics or general relativity, is useless in this context. Existing recurrent spiking neural networks (SNNs) cannot be trained on complex tasks or simulated at scale, primarily due to a lack of computational efficiency in digital processing. Because all neurons inherently work in parallel, they essentially represent up to 100 billion processors (for example, let's assume the average human brain has 100 billion neurons). Replicating this processing power is possible if each neuron represents only 1 floating-point of computational power, but in reality, the accuracy of neuron simulation required to achieve the Turing test is unknown: if each neuron is on the order of 100 Mflops, it is impossible to achieve with current technology. In reality, the best possible computational power is 1E16 Flops, and 100E9 x 100E6 = 1E11 x 1E8 = 1E19, which is approximately 1,000 times greater than the computing power of the world's best high-performance computing systems. Even non-real-time simulations would be too slow to obtain any meaningful results in a reasonable amount of time. Biological brains are also orders of magnitude more computationally efficient than digital computers: for example, a brain typically consumes 20W for 100 billion neurons (5 billion neurons / W), while digital simulations would require several orders of magnitude more power even for simple neuron models such as integrating and firing spiking neurons.

[0007] Therefore, a new alternative approach is to use biological neural networks instead of silicon-based digital computing to replicate high-level cognitive processes. Recent advances in biotechnology now facilitate the cultivation and assembly of biological neural networks from embryonic stem cells, such as rat embryonic stem cells, and from differentiated human induced pluripotent stem cells (IPSCs). C). The cultured BNNs can then be stimulated and read out using multi-electrode arrays (MEAs). So far, the main use of MEAs has been in the development of neuronal and brain models for pharmacology, drug testing and toxicology studies, as well as for a better understanding of common brain diseases such as Alzheimer's and Parkinson's. Other industrial applications have also been proposed in the past few years. The application of BNNs in aircraft control was proposed, for example, by DeMarse et al. in "Adaptive flight control with living neuronal networks on microelectrode arrays" in the proceedings of the 2005 IEEE International Joint Conference on Neural Networks. US Navy US Patent 7947626 discloses neural network MEAs derived from cultured progenitor cells, which can be used to detect and / or quantify various biological or chemical toxins. On the same track, Koniku (www.koniku.com) was the first company to develop a wet chip based on a computing chip of cultured neural cells and has been commercializing highly specialized products for the detection of odor compounds in the security, military and agricultural / food markets since 2017. As described in their patent application WO 2018 / 081657, neuronal cells can be modified by various biotechnological processes (e.g. gene editing, methylation editing, etc.) to express a unique spectrum of odorant receptors with cell surface receptors, as is known in the biological state of the art. The neuronal cells can be interfaced with a computer through state-of-the-art neurophysiological interfaces such as MEA (multi-electrode array) electrodes. The computer can then measure the electrical signals generated by the nerve cells when exposed to odorant compounds in a dedicated chamber and detect the presence of certain odorant compounds by conventional signal processing methods, which may include artificial neural networks (ANNs) and machine learning classifiers for training and classification. ANNs are particularly useful for distinguishing multiple signals when networks of different neurons with different odorant receptor properties are used in real environments that may involve complex combinations of multiple detectable compounds. A major limitation of this approach is that it is limited to very specific sensory applications.

[0008] Baker Hughes' U.S. patent application US20140279772 discloses the use of a cultured biological neural network in an apparatus for processing downhole signals, which are transmitted into the formation through a borehole in a container. The biological neural network is connected to MEA electrodes to receive input signals from sensors and output measurements outside the neural network. The proposed BNN system also includes environmental control components, such as nutrient dispensers. Although the present disclosure mentions the advantages of BNN over traditional computing systems in challenging environments in terms of parallel computing capabilities, robustness to vibration and electrical noise, and self-healing capabilities, it does not clearly explain how to manage the system over time, especially the dynamic learning process. It is very specialized for specific applications, so preprocessing learning can be applied as an offline preparation process rather than in a challenging runtime environment.

[0009] In his 2013 doctoral dissertation titled "Simulating and Cultivating Neural Network Computation," Ju Han from the University of Singapore explored the capabilities of dissociated neuronal cultures derived from 18-day-old rat embryonic cortical cells within the context of a state-dependent computing paradigm. His research demonstrated that random networks composed of living neurons can process complex spatiotemporal information, making them suitable for implementing prototype neural computer devices based on the LSM (Liquid State Machine) paradigm. To test the neuronal cultures' ability to classify temporal and complex spatiotemporal patterns, he designed two stimuli: a jittered spike-trained template classification task (a benchmark for LSM) and a randomized piano music classification task. Processing temporal input depends on memory decay, and Ju Han observed that short-term memory in his dissociated neuronal cultures could exceed four seconds in his setup. To control the neuronal cultures, he employed optogenetics (light stimulation of genetically modified neurons) combined with multi-electrode arrays (MEAs) for both stimulating neuronal input and recording neuronal output. While MEAs can provide electrical stimulation (BNN input) and measurement (BNN output) at low spatial and temporal resolution, optogenetics allows for more precise stimulation control, contactless manipulation, and repeated interrogation of neurons, and as a result, this field has further developed into an intensive research area over the past five years. Examples of recent advances in optogenetic stimulation can be found in the following papers: Agus and Janovjak, “Optogenetic Methods in Drug Screening: Techniques and Applications,” in Curr. Opin., Biotechnol., December 2017, and Barral and Reyes, “Optogenetic Stimulation and Recording of Primary Cultured Neurons with Spatiotemporal Control,” in Bio Protoc, August 2017, which describe a fast video projector based on the working principle of a digital micromirror device (DMD) that is capable of spatially focusing light stimulation to single neurons while achieving temporal display patterns at 1.44 kHz and above.

[0010] To optimize the overall functionality of the BNN system, Ju Han proposed further controlling the biological culture unit with standard machine learning methods (specifically genetic algorithms) to define electrical stimulation and processing, ultimately achieving advanced functions. Instead of single neuron readouts, which are often considered in neuroscience experiments, the network layer outputs can be read and processed as multivariate signals (for example, using MEA electrophysiological measurement probes in combination with signal processing software such as Matlab).

[0011] Further research in neuroscience and neural computing emphasizes that biological neuronal systems require learning processes, just as the brain develops its computational capabilities. In his 2013 doctoral dissertation, Ju Han demonstrated that biological neuronal systems are capable of learning or being designed, possibly through light stimulation. This learning ability of the network, combined with drug manipulation, forms a possible step toward optimizing neural circuits for computation. BNN systems can be trained to produce behaviors specified by a reference model through reinforcement learning, such as by releasing global reward or punishment signals based on feedback from behavioral outcomes. For example, Ju Han proposed using NDMA receptor antagonists as a therapeutic approach to reinforcement learning.

[0012] More generally, in addition to open-loop systems, which are still widely used in neuroscience experiments, closed-loop systems can also be designed as in conventional automated systems engineering. Current state-of-the-art BNN systems are tailored for small-scale, very specific applications, and therefore have two main limitations when used as single components in overall general-purpose wetware computing systems with adaptive (different needs) and evolving (over time) advanced cognitive capabilities: their architecture is limited to a single input (single neuron or layer stimulation using predefined settings and protocols) and output layer design (single neuron or layer measurement using predefined biological or signal processing methods); their inherent network connectivity, once with or without preliminary training, can only process one function at a time.

[0013] Therefore, new solutions and architectures are needed to exploit the advanced cognitive processing capabilities inherent in biological neural networks as a low-power alternative or co-processing complement to more traditional silicon-based information technology processing systems, devices, software, and electronic chips. Summary of the Invention

[0014] An automated processing system for transforming a spatiotemporal input data signal into a spatiotemporal output data signal is described, the system comprising: an in vitro biological culture of neural cells (BNN core unit); an input stimulation unit (SU) adapted to apply an input spatiotemporal stimulation signal to a first group of neural cells; an output readout unit (RU) adapted to capture an output spatiotemporal readout signal from a second group of neural cells; one or more nutrient tanks connected to one or more nutrient dispensers for injecting one or more nutrients into the biological neural cell culture; one or more additive tanks, each additive tank connected to one or more additive dispensers for injecting one or more additives into the BNN culture; one or more nutrient waste collectors for filtering and discharging nutrient waste from the BNN culture; and one or more additive waste collectors for filtering and discharging nutrient waste from the BNN culture. filtering and discharging additive waste; one or more vascularized networks connecting the nutrient dispenser; the additive dispenser, nutrient waste collector and additive waste collector of the BNN culture; one or more sensors for measuring at least one environmental parameter of the BNN culture; and an automated controller configured to adapt the stimulation signal to the input data signal, adapt the output data signal to the readout signal, and control at least one of the BNN core unit environmental parameters, BNN neural cell culture nutrient supply, BNN neural cell culture additive supply, BNN neural cell culture nutrient waste collection, and BNN neural cell culture additive waste collection, thereby maintaining the homeostasis of the BNN neural cell culture over time, so that the spatiotemporal input data signal is continuously converted into the spatiotemporal output data signal.

[0015] The automated controller may include: a preprocessing unit that transforms the input data signal into the stimulation signal using at least one of a spatiotemporal signal filter, a spatiotemporal signal classifier, a machine learning algorithm based on a mathematical or statistical model, an artificial neural network, a convolutional neural network, a support vector machine classifier, a random forest classifier, a genetic algorithm, a genetic programming algorithm, or a reservoir calculation method.

[0016] The automated controller may include a post-processing unit for transforming the readout signal into the output data signal using at least one of a spatiotemporal signal filter, a spatiotemporal signal classifier, a machine learning algorithm based on a mathematical or statistical model, an artificial neural network, a convolutional neural network, a support vector machine classifier, a random forest classifier, a genetic algorithm, a genetic programming algorithm, or a reservoir computation method.

[0017] Also described is a method for converting a spatiotemporal input data signal into a spatiotemporal output data signal using an automated controller and a biological neural network (BNN) core unit, wherein the BNN core unit includes at least an in vitro culture of neural cells, and the in vitro culture of neural cells is suitable for feeding a stimulation spatiotemporal signal to a first group of neural cells using an input stimulation unit (SU), and is suitable for reading out a spatiotemporal signal from a second group of neural cells using an output readout unit (RU). The method includes: pre-processing the spatiotemporal input data signal using the automated controller to form the stimulation spatiotemporal signal; post-processing the readout spatiotemporal signal using the automated controller to form the spatiotemporal output data signal; and controlling at least one of the BNN core unit environmental parameters, the BNN neural cell culture nutrient supply, the BNN neural cell culture additive supply, the BNN neural cell culture nutrient waste collection, the BNN neural cell culture additive waste collection, pre-processing parameters or post-processing parameters, thereby maintaining the homeostasis of the BNN neural cell culture over time, so that the spatiotemporal input data signal is continuously converted into the spatiotemporal output data signal.

[0018] The method may also include minimizing the error between the output data signal and the target output data signal by adjusting at least one of the BNN core unit environmental parameters, the BNN neural cell culture nutrient supply, the BNN neural cell culture additive supply, the BNN neural cell culture nutrient waste collection, the BNN neural cell culture additive waste collection, preprocessing parameters or post-processing parameters. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 The core BNN unit is described as the building block of a biological computational server.

[0020] Figure 2a ), Figure 2b ), Figure 2c ), Figure 2d ), Figure 2e ) propose different arrangements and fabrication schemes to assemble core BNN units.

[0021] Figure 3 An exemplary schematic diagram of the automated vascularization system (AVS) operating using the BNN core unit is described.

[0022] Figure 4 and Figure 5 Shown are side cross-sectional views of two exemplary sponge-like structures as possible hosts for neural cells with intrinsic vascularized network support for growth in 3DBNN culture.

[0023] Figure 6A schematic diagram showing a possible automated BNN growth, maintenance, and control system.

[0024] Figure 7 Another schematic diagram of possible automated BNN growth, maintenance, and control as a biological operating system (BOS) is shown.

[0025] Figure 8 A learning process that can be implemented by real-time processing software is shown.

[0026] Figure 9a )and Figure 9b ) shows a possible embodiment of the BNN Biocomputing Stack (BCS).

[0027] Figure 10a )and Figure 10b ) shows a possible embodiment of a BNN biological computing stack (BCS) (T-BCS) suitable for training.

[0028] Figure 11 An exemplary T-BCS implementation with an ANN is shown.

[0029] Figure 12 A possible maintenance procedure of T-BCS is shown.

[0030] Figure 13 A host server operating using the T-BCS wet software architecture is shown.

[0031] Figure 14 Describes possible functions that may be operated by a master server to manage user clients.

[0032] Figure 15 A general architecture with load balancing is also described for serving multiple clients from the same T-BCS server host.

[0033] Figure 16 Possible applications of the proposed BNN server for image processing are compared with applications with deep learning architectures.

[0034] Figure 17 It shows that rat cortical stem cells have a neurosphere with a width of about 200 meters after 4 days of maturation.

[0035] Figure 18 、 Figure 19 、 Figure 20 Schematic showing examples of 12 electrodes regularly distributed along the periphery of a neurosphere of cortical neural stem cells at three different growth stages.

[0036] Figure 21 A photograph showing a biological neural network on an MEA circuit.

[0037] Figure 22 Adherent cells and Microscope images of the stroma. DETAILED DESCRIPTION

[0038] BNN core unit

[0039] Figure 1 The core BNN unit 100 is described as a building block of a biological computing server. The biological material in the BNN unit is typically composed of, but not limited to, an active biological culture 120, which is typically a combination of multiple living neural and glial cells. The cells can be assembled by a variety of different processes, such as, but not limited to, cell culture or organogenesis-like methods. Throughout this disclosure, the term cell culture is used indiscriminately to refer to in vitro cell growth and life maintenance outside of its natural in vivo environment. The BNN unit can be arranged in 2D or 3D. The stimulation unit 110 (SU) represents the input interface between the digital data input signal 105 and the biological culture 120. It is used to selectively stimulate different neurons, dendrites or axons. The stimulation unit 110 can control the digital data input signal transferred to the biological culture by spatially (addressing different neurons), temporally (stimulating with time-varying signals) and / or spatially-temporally changing the digital data input signal transferred to the biological culture. Examples of implementations of SU 110 include, but are not limited to, multi-electrode arrays (MEAs), patch clamps, light-induced stimulation (such as optogenetic systems), magnetic or electric fields, ion stimulation, focused lasers, optical tweezers, or mechanically induced stimulation via gravity or pressure changes. Readout unit 130 represents the output interface between biological culture 120 and digital data output signal 135. It is used to selectively measure the activity of different neurons, dendrites, or axons. Readout unit 130 can control the conversion of biological culture activity into digital data output signals (which may be multidimensional) by sampling them spatially (capturing the individual activity of different neurons), temporally (capturing time-varying signals), and / or spatiotemporally. Examples of implementations of RU 130 include, but are not limited to, multi-electrode arrays (MEAs), patch clamps, imaging systems, ion-sensitive sensors, electrically or magnetically sensitive sensors, chemical sensors, and other sensors suitable for neuronal cultures.

[0040] A subset of neurons can be stimulated with input signals via a BNN input interface such as an electrical signal from a multi-electrode array (MEA). Alternatively, a subset of neurons can be genetically modified to receive optical stimulation from an optogenetic system as an input signal. Using a measurement system such as a multi-electrode array (MEA) that receives electrical signals, the electrical activity of neuronal cells can be monitored at several locations in a biomaterial as the output of the BNN unit. Alternatively, a subset of neurons can be genetically modified to express fluorescence as an output signal from the BNN to an imaging sensor system. However, there are many other systems that can implement or facilitate the connection of BNNs, such as processes that utilize similar concepts to those in the human body, such as processes that convert electrochemical stimulation into mechanical motion and are observable in muscle movement or speech.

[0041] In a possible embodiment, the Maxwell Biosystems ( https: / / www.mxwbio.com ) can be used as a host platform for the BNN core unit 100. Isolated cell culture BNNs 120 can be plated and grown on the MaxOne microsensors in CMOS technology according to the protocol provided at the following website: https: / / www.mxwbio.com / applications / neuronal-networks / applications / neuronal-ne tworks / :

[0042] Sample cell culture plating procedure

[0043] ○ The electrode array surface was pre-coated with a thin layer of 0.05 wt% poly(ethyleneimine) (PEI) (Sigma, MO, USA) in 8.5 pH borate buffer (Chemie Brunschwig, Basel, Switzerland).

[0044] ○ A drop of 0.02 mg / ml laminin (Sigma) was added to the culture medium (Invitrogen, California, USA) for cell adhesion.

[0045] ○ The seed cell suspension was dripped onto the array in 6-1 drops.

[0046] ○ After 20-30 min add 1 ml of plating medium.

[0047] ○ After 24 hours, the plate medium was replaced with 1-2 ml of growth medium and the cultures were maintained in an incubator with controlled environmental conditions (37°C, 65% humidity, 5% carbon dioxide).

[0048] ○ Change 50% of the growth medium twice a week.

[0049] In the protocol proposed by Maxwell Biosystems, the plating medium may consist of 850 ml of culture medium supplemented with 10% horse serum (HyClone, UT, USA), 0.5 mM GlutaMAX (Invitrogen, CA, USA) and 2% B27 (Invitrogen, CA, USA), but it will be apparent to those skilled in the art of cell culture that other formulations are possible.

[0050] In the protocol proposed by Maxwell Biosystems, the growth medium may consist of 850 ml DMEM (Invitrogen, CA, USA) supplemented with 10% horse serum, 0.5 mM GlutaMAX and 1 mM sodium pyruvate (Invitrogen, CA, USA), but it will be apparent to those skilled in the art of cell culture that other formulations are possible.

[0051] In a possible embodiment, a Maxwell MEA microsensor may be operated as a stimulation unit 110. The Maxwell MEA microsensor is capable of stimulating BNN activity from an input digital data pattern using a subset of active stimulation electrode sites. The input digital data pattern 105 may then be prepared by various data processing methods and software. In a possible embodiment, a Maxwell stimulation module may be used as an SU 110 to provide 32 stimulation channels. Each stimulation channel can provide a voltage of up to ±1.6V or a current amplitude of ±1.5mA with an amplitude resolution of 2nA and a time resolution of 2s. The MaxLab Live software component can generate a variety of digital data stimulation patterns suitable for these resolutions, such as monophasic, biphasic, triphasic pulses, ramp waveforms, and other custom pulse shapes.

[0052] In a possible embodiment, a Maxwell MEA microsensor can operate as a readout unit 130. The Maxwell MEA microsensor is capable of outputting digital data readouts of BNN activity that can be recorded simultaneously using multiple active electrode sites on a configurable time scale from microseconds to months. The digital data readouts 135 can then be processed by various signal processing methods. The readouts can also be visualized in an imaging system, for example as a raster plot. Although the current Maxwell Biosystems MEA technology embeds a high-resolution CMOS-based microelectrode array as a two-dimensional electroplating, as described in "A 1024-Channel CMOS Microelectrode Array with 26,400 Electrodes for In Vitro Recording and Stimulation of Electrogenic Cells," IEEE Journal of Solid-State Circuits, Vol. 49, No. 11, pp. 2705-2719, 2014, other arrangements are possible.

[0053] Although the above possible embodiments have been described using the Maxwell Biosystems MEA solution as an exemplary implementation of a core BNN unit with high density and high throughput based on recent technological advances, it will be apparent to those skilled in the art that other neurotechnologies, electrophysiological and / or optogenetic systems, circuits, devices, probes, components, software, protocols and methods may also be used alone or in combination with each other to provide a functional core BNN unit 100, for example, the Maxwell Biosystems MEA solution from Harvard Biosciences www.multichannelsystems.com )、3Brain( www.3brain.com ), Nuvectra's subsidiary NeuroNexus ( www.neuronexus.com )、AxionBiosystems( https: / / www.axionbiosystems.com ), Charles River Laboratories( www.criver.com ), Plexon( www.plexon.com )、by Koniku( www.koniku.com ), Georgia Tech Porter Laboratory ( https: / / sites.google.com / site / neurorighter / ), mesh electronics for chronic recording at the single-neuron level from Harvard University's Lieber Lab ( http: / / cml.harvard.edu ) and other products developed.

[0054] It will be apparent to one skilled in the art that different types of neuronal cells may be employed as the biological basis of the BNN 120. In addition, the biological basis may of course comprise a single neural cell type, or a predetermined combination of different neural cell types or even other cells. Furthermore, the cell type composition may vary throughout the 2D or 3D structure of the BNN 120. In a possible embodiment, rat embryonic neural stem cells (NSCs), such as those from ThermoFisher Scientific, provided by Invitrogen, may be used. Cell line catalog numbers N7744-100, N7744-200. In an alternative embodiment, human neural stem cells (hmNPCs) such as those from Lonza Poietics TM Neural progenitor cells (NHNP), MiliporeSigma or ThermoFisher Scientific StemPro™ neural stem cells, etc. Neuronal cells can be maintained in vitro using biological culture media such as MEM (Modified Eagle's Medium Gibco) or DMEM (Dulbecco's Modified Eagle's Medium - Gibco, Invitrogen, ThermoFisher).

[0055] In possible embodiments, the BNN 120 may be in the form of a neurosphere or neurovolume system, an adherent monolayer system or other specialized arrangements and configurations for immobilizing BNN cells and / or other cells such as neural stem cells (NSCs). It will be apparent to those skilled in the art of biomaterials that the BNN 120 may be arranged in 2D or, preferably, in 3D with various scaffold types to provide a suitable living environment for the cells. This includes, but is not limited to, processes such as growing into brain organoids or simply maintaining the location of living cells. The resulting in vitro brain organoids can operate in a sustainable manner that is as similar as possible to the mammalian brain environment. Examples of state-of-the-art scaffold forms and materials, such as hydrogels, which can be used for the three-dimensional culture and differentiation of various rat, mouse and human neural cells are shown in Table 1 of "Three-dimensional in vitro culture scaffolds for neural lineage cells" by A. Murphy et al. in Acta Biomaterialiae Sinica, 54 (2017) 1-20. Hybrid hydrogels may also be used, such as those recently described by Mauri et al. in Biomater Sci.2018 Feb 27;6(3):501-510. [The text then abruptly shifts topics.] ...

[0056] Biocompatible materials may also be particularly suitable for the microelectronic components of the stimulation unit SU 110 and / or the readout unit RU 130. As a possible embodiment, Figure 2a ) presents a possible application of mask etching to create a biocompatible layer 251 that closely matches the underlying multi-electrode array structure 250. In effect, without precise positioning, the readout electrode or stimulation electrode simultaneously reads / injects signals into multiple neurons. Etching using a mask 252 opposite the MEA array structure 250 can be used to create a biolayer 251 that induces neuron positioning at a more precise location by aligning the biocompatible material 251 with the underlying MEA 250, so that the stimulation unit 110 can stimulate the BNN 120 at the neuron level and / or the readout unit 130 can read the BNN 120 at the neuron level.

[0057] In a possible embodiment, an adhesion layer may be specifically applied at the locations where the RU sensors and / or SU probes are located so that neural cells preferentially adhere and / or grow onto these areas, thereby facilitating their control through the RU and SU interface. As a possible alternative embodiment, Figure 2b ) shows a schematic diagram of a BNN core unit assembly support 270 suitable for ensuring maximum efficiency of communication between the RU and SU interfaces and the neural cells. In conventional methods, the RU and SU interface locations 271 can be coated with an adhesion layer 272, and an additional layer 274 with masking properties can be further deposited outside the RU and SU interface locations to prevent the unit 273 from adhering to and unfolding outside the control of the RU and SU interfaces. The interface layer can also be composed of special membranes that allow optimal adhesion and exchange of molecules and atoms for different purposes, including, for example A matrix or any similar material that simulates growth and connection.

[0058] Figure 2c ) shows a side view of a possible embodiment of a 3D layered stack of neurospheres 280. Nutrient and additive liquid solutions useful for growing and maintaining BNN cultures can optimally flow through the entire neurosphere stack joint. Flow can be facilitated and controlled by various means, such as natural gravity, centrifugation, electrical or magnetic forces, and / or pumping.

[0059] Figure 2d) shows a possible embodiment of an optogenetic stimulation unit SU interface for generating spikes on transgenic photosensitive neurons 293. A number of light beams 290, 291, and 292 can be aimed at the same neuron in such a way that all the light beams cross at the location of the nerve cell 293, which in turn generates spikes. This method is capable of generating spikes in a three-dimensional aggregate of nerve cells. Let T be the intensity threshold of the light stimulation, above which the illuminated neuron emits a spike signal. Then the intensity values I of the p light beams can be defined p , such that sum(I p ) > T, even though the individual intensity of each light beam is lower than the spike threshold I p < T. It is obvious to those skilled in the art of optogenetics that various illumination devices can be used according to the characteristics of transgenic photosensitive nerve cells. In a possible embodiment, a laser for optogenetics can be used as the illumination beam. To stimulate a specific neuron at a given depth within a 3D aggregate of neurons, the laser (or multi-spectral) beam can be focused so as to precisely reach the maximum energy of light at the neuron depth position. The target position can also be adjusted by changing the position of the light-emitting device and / or by adjusting the orientation of the mirror used to deflect the direction of the light beam.

[0060] As a possible alternative embodiment, Figure 2e ) shows an alternative embodiment in which the neuron 260 is deposited on a digital display support 261. The digital display support can be, for example, an LCD screen or an OLED screen. Current display technologies, such as the screens developed for the smartphone industry, typically employ pixel sizes on the order of dozens of micrometers. In addition, other technologies such as LCoS (liquid crystal on silicon), DLP, or DMD micromirror methods are less than 5 μm, while neuron sizes range from 4 to 100 μm, so it is possible to selectively control pixel illumination outside the neuron 262, at the boundary 263 of the neuron, or in the line 264 with the neuron. In addition, color pixels can be selected to obtain a specific effect on neuron activity according to the characteristics of transgenic photosensitive nerve cells.

[0061] Automated BNN fabrication, growth, and maintenance

[0062] To develop high-level cognitive abilities, BNN core cells need to be maintained in a stable, sustainable, and safe environment for a sufficient period of time to optimize the growth and training of their internal biological networks according to their biological needs, just as in vivo. As is readily apparent to those skilled in the biotechnology field, various automated systems have been developed for efficient biomanufacturing using well-established cell lines for many biotechnology industrial applications; however, mammalian neural cells are particularly fragile when handled in vitro outside of their natural brain environment, which provides exceptional protection from potentially disruptive external factors such as mechanical strain or shock, contamination from pathogens or chemicals, and exposure to light radiation. Another issue is that traditional cell line cultures typically operate at the cellular level (e.g., cells are the basic factory components for producing certain proteins of interest), whereas for neural cells, they need to operate at a higher dimensional level, often through interconnections in multidimensional networks. Another problem is that the human brain takes years to express its high-level cognitive potential in vivo, and there is currently no research evidence suggesting that this process could be accelerated in vitro (Pascal, “The Rise of Three-Dimensional Human Brain Cultures,” Nature, Vol. 553, January 2018).

[0063] Therefore, a dedicated biological neural network automation method and system is needed to manufacture, maintain and control neural cell cultures in a sustainable and repeatable process that can be run automatically and in parallel. The manufacturing process includes the assembly of BNN core unit elements and the growth of BNN cultures into functional BNN core units, including automated learning processes. The maintenance and control process includes various activities such as feeding, stimulation, measurement, as well as curing and cleaning. Possible embodiments of such an automated method and system will now be described in more detail.

[0064] To expand BNN core unit development beyond the current manual laboratory setup, the manufacturing process could include automated assembly of BNN core unit components. In a possible embodiment, 3D bioprinting could be employed to combine neural cells with a scaffold and the necessary components for their growth. Examples of 3D biocompatible tissue engineering include products developed by Sichuan Revotek Co., Ltd., Biosynsphere, Organovo, Aspect Biosystems, and others. In one possible embodiment, BNN core unit components could utilize stem cell-derived neural progenitor cells, as described in "3D-Printed Stem Cell-Derived Neural Progenitor Cells for Spinal Cord Scaffold Generation" by Joung et al. in Adv. Funct. Mater. 2018. This facilitates optimal cell positioning directly on the scaffold. Cell clusters in bioinks, such as cell-laden hydrogels, can be deposited in continuous channel layers at a resolution of approximately 200 μm. As the authors report in their research on tissue models and future spinal cord injury implants, this approach facilitates axonal proliferation and maintains cell viability and mechanical stability during in vitro CNS tissue construction.

[0065] In order to automatically maintain the pre-assembled BNN core units in operation in addition to the current manual laboratory maintenance tasks, the second stage of the automated process could include automatic vascularization through microfluidic circuits to biologically supply the BNN core unit cells and collect their biological waste. So far, only limited solutions have been proposed for the in vitro vascularization of 3D brain organoids, and therefore their size remains limited due to the central necrosis of the cells, which can no longer be reached by peripheral nutrients after the organoids reach a certain depth and density. Therefore, research biologists have recently proposed to transplant them onto the adult mouse brain (Mansour et al., "In vivo model of functional and vascularized human brain organoids", Nature Biotechnology, Volume 36, Issue 5, May 2018 and Lancaster, "Vascularization of brain organoids", Nature Biotechnology, Volume 36, Issue 5, May 2018). The technical feasibility of vascularizing brain organoids with the patient's own endothelial cells was also recently demonstrated in Pham et al. "Generation of Human Vascularized Brain Organoids" (Neurological Reports, Vol. 29, No. 7, pp. 588-593, May 2018). However, the latter approach is inherently limited to natural jet distribution and can only be controlled from the outside in a black box manner, which limits the automation capabilities to very simple models. In order to overcome this limitation of prior art BNN cultures, various possible alternative embodiments can be considered, either individually or in combination, as will now be described in further detail.

[0066] Figure 3A schematic diagram of a first possible embodiment of an automated vascularization system (AVS) 300 operating with the BNN core unit 100 is provided, comprising:

[0067] One or more nutrient tanks 319 connected to one or more nutrient dispensers 320 to inject one or more nutrients into the BNN culture 120. The nutrient dispensers 320 may include a mixer-injector active module that is responsible for delivering the correct dosage of nutrients based on the specific nutrient requirements of the BNN culture using a mechanical device 351 such as a valve, syringe, or pump.

[0068] One or more additive tanks 321, 323 are each connected to one or more additive dispensers 322, 324 to inject one or more additives into the BNN culture 120. The additive dispensers 322, 324 may include a mixer-injector active module that is responsible for delivering the correct dosage of additives based on the growth requirements of the BNN culture using a mechanical device 352, 353 such as a valve, syringe, or pump.

[0069] One or more nutrient waste collectors 325 having a mechanism 331, such as a valve, syringe, or pump, for filtering and draining the nutrient waste into a tank 335 and / or back into the BNN.

[0070] One or more additive waste collectors 326, 327 having mechanisms 332, 333, such as valves, syringes, or pumps, for filtering and draining additive waste into tanks 336, 337 and / or back into the BNN.

[0071] One or more angiogenesis networks connect nutrient and additive distributors 320, 322, 324 and nutrient and additive waste collectors 325, 326, 327 to the BNN culture 120. Different angiogenesis streams can also be reinjected into the system to minimize losses.

[0072] Nutrients may include amino acids, carbohydrates, vitamins, and minerals, which may be the same or in separate liquid solutions.

[0073] Additives may include chemicals, drugs, or other elements, such as dopaminergic stimulation enhancers that increase the dopaminergic response of BNNs, and dopaminergic stimulation inhibitors that reduce the dopaminergic response of BNNs. It will be apparent to those skilled in the art of automation that this enables the reproduction and control of dopaminergic / anti-dopaminergic systems. More generally, additives may include chemicals known to affect neurotransmitters in the central nervous system, such as botulinum toxin, nicotine, curry, amphetamines, cocaine, MDMA, strychnine, THC, caffeine, benzodiazepines, barbiturates, alcohol, opioids, and the like. Additives may also include growth factors, hormones, and gases (such as carbon dioxide, oxygen, and the like).

[0074] Each nutrient dispenser 320 is interconnected with the BNN culture 120 via a nutrient vascularization network 328 that delivers nutrients to the BNN cells. Each additive dispenser 322, 324 is interconnected with the BNN culture 120 via an additive vascularization network 329, 330 that delivers additives to the BNN cells. The nutrient vascularization network 328 and the additional vascularization networks 329, 330 can be the same or different networks.

[0075] Each nutrient waste collector is interconnected with the BNN cells 120 through a nutrient waste vascularization network that transports nutrient waste from the BNN cells. Each additive waste collector is interconnected with the BNN culture 120 through an additive waste vascularization network that transports additive waste from the BNN cells.

[0076] The nutrient waste vascularization network can be the same or a different network than the nutrient vascularization network 328. The additional waste vascularization network can be the same or a different network than the additional vascularization networks 329, 330. The vascularization networks can be fabricated using 3D bioprinting using biocompatible materials or grown from stem cells on the BNN 120 culture support.

[0077] It is important to point out that vascularization can be considered as a similar system in the human brain or as a completely different type of system, such as the structure of porous materials. Figure 4A side cross-sectional view of an exemplary embodiment of a sponge-like structure as a possible host for neural cells 440 having an inherent vascularized network support to enable them to grow in a 3D BNN 120 is shown. The sponge-like material 439 is preferably soft and compressible to mechanically accommodate the BNN cells 440 (as shown in 410) in its pores 442, while subsequently adjusting to accommodate their growth and development as a 3D BNN culture. The sponge-like material 439 is further preferably porous and absorbent to transport nutrients and / or additives and / or waste throughout the BNN culture. In a possible embodiment, a solution containing the neural cells 440 can be soaked in the sponge-like structure 439 so that the neural cells are regularly distributed in its pores and channels. In a possible embodiment, a polyethyleneimine (PEI) polymer can be used to increase adhesion factors. In another possible embodiment, the neural stem cells 440 can be soaked in a solution containing growth factors (e.g., fibroblast growth factor FGF2) to prevent cell specialization until the stem cells are distributed into the sponge-like structure 439. Once the distributed cells have adhered to the sponge 439, the BNN culture 120 is assembled and the sponge 439 can be further used as its vascularization system. The adhesion and growth factor solution can be washed out and replaced with another nutrient and additive solution 441 suitable for the specialization, growth and maintenance of the BNN cultured cells. In an alternative embodiment (not shown), induced pluripotent stem cells can be mechanically inserted and cultured directly onto the sponge 339 as its inherent vascularization network system.

[0078] Figure 5 A top cross-sectional view of another exemplary embodiment of a sponge-like structure as a possible host for neural cells 544 with an inherent vascularized network to support their growth within a 3D BNN culture while facilitating interfacing the electrical and / or optical components of a stimulation unit and / or readout unit with the 3D BNN culture is shown. To this end, the sponge-like structure can be shaped into a sphere, with electrophysiology clamps or electrodes passing through the sponge-like structure from various locations throughout the sphere and penetrating the sponge-like structure at varying depths.

[0079] In another possible exemplary embodiment (not shown), 3D BNNs can be cultured in a sandwich-like structure along biocompatible filaments suspended between two planes. The neural cells can be attached to the biocompatible filaments in various ways. In one possible embodiment, the filaments can be coated with an adhesion factor such as PEI, but other embodiments are also possible.

[0080] BNN controller

[0081] Figure 6A schematic diagram of a possible automated BNN growth, maintenance, and control system is shown, comprising a BNN core unit 120, a stimulation module 110, a readout module 130, and an automated angiogenesis system (AVS) 300, which may include nutrient and additive tanks and dispensers, and a waste collector. In a preferred embodiment, the operation of the AVS is coordinated by a BNN automation controller 600, which is responsible for calculating in real time the selection and amount of nutrients and additives to be injected into the BNN culture by the AVS, as well as the amount of waste to be collected from the BNN culture by the AVS. The BNN automation controller can be a computer processor constructed with electronic hardware and suitable for executing software algorithms. In a simple embodiment, the BNN automation controller can operate in an open loop, identifying and quantifying the necessary nutrients and additives, and deriving the resulting waste values ​​based on knowledge from scientific expertise in neurophysiological parameterization. In another embodiment, the BNN automation controller can further continuously monitor the readout information 635 from the BNN readout module 130 to determine the latest BNN culture status and can adjust the AVS parameters accordingly.

[0082] In another embodiment, the BNN automation controller can control the stimulation signal 605 so that the stimulation unit operates synchronously with the BNN state. For example, the nutrients and additives injected by the AVS into the BNN culture medium and the waste collected vary depending on the maturity stage of the BNN cell life cycle: differentiation stage (assembly), growth stage (learning), operation stage (stable function) and death stage (production of additional waste). The proportion of nutrients or drugs injected or processed can be controlled by a nutrient dispenser, an additive dispenser and their respective waste collectors. Through these feedback loops, the amount of nutrients and drugs can be dynamically adjusted to: 1 / optimize lifespan, 2 / minimize manual care, 3 / adjust responsiveness, and stabilize the performance of the BNN. Therefore, the homeostasis of the BNN can be maintained over time.

[0083] In a possible embodiment, the BNN automation controller 600 can directly feed the raw data input signal 105 as the stimulus signal 605. In an alternative embodiment, the BNN automation controller can further include a pre-processing unit 610 responsible for converting the data input signal 105 into the stimulus signal 605. This enables the BNN automation controller to better adapt the raw input to the actual stimulus unit (SU) capabilities and the BNN capabilities, thereby optimally performing the end-to-end BNN processing task without requiring the end user to specifically adapt his / her input signal to every possible configuration. This also facilitates learning tasks, as the capabilities of the BNN can evolve over time, and the signal pre-processing 610 can therefore also adapt accordingly.

[0084] In a possible embodiment, the BNN automation controller can directly output the raw BNN readout signal 635 as the signal processed from the end-to-end BNN system. In an alternative embodiment, the BNN automation controller 600 can also include a post-processing unit 630 responsible for converting the raw BNN readout signal 635 into the output signal 135. This allows the BNN automation controller to better adapt the raw readouts to the needs of real-world applications, which may be too noisy to be easily interpreted. One exemplary application is to extract relevant signals from a stream of spikes using a spike sorting signal processing algorithm, but other approaches such as statistical models, classifiers, or even machine learning methods can also be used. This also facilitates learning tasks, as the end-to-end BNN system can be trained to correlate the composite target output signal 135 to predefined input signals 105, without requiring the end user to specifically interpret the measured output signal for each possible internal BNN biological culture or readout unit configuration. This also facilitates learning tasks, as the capabilities of the BNN can evolve over time, and thus the signal post-processing 630 can adapt accordingly.

[0085] Figure 7Another schematic diagram of possible automated BNN growth, maintenance and control as a biological operating system (BOS, by analogy to a computer operating system) is shown. Similar to a traditional computer operating system or cloud computing service that abstracts the underlying hardware-specific devices and operations, such a BOS can host real-time functional control software for end users to perform BNN processing unit functions by abstracting the BNN automation process so that the end user does not notice them. The proposed BOS operates in conjunction with the BNN core unit culture 120, the stimulation module and readout module SU / RU 110 / 130, and the BNN health control unit 710, as previously described for an automated angiogenesis system (AVS). The BNN health control unit 710 may include a chemical control unit with nutrient and additive tanks and dispensers and a waste collector, as well as an environmental control unit responsible for controlling other BNN culture environment parameters (such as temperature, pressure, humidity, oxygen or carbon dioxide ratios) and other parameters. It may include, for example, digitally controlled micropumps for delivering chemicals and temperature, humidity, pH or carbon dioxide sensors, or it may even be able to change any other environmental parameter, including, for example, sound waves or light waves in any frequency band. The BNN health control unit is responsible for monitoring the health of the BNN cells under real-time supervision by the BNN health control software 700. Therefore, the health monitoring system operates as a steady-state system responsible for regulating BNN performance over time, as BNN performance can naturally drift over time, transparently to the end user. It operates by adjusting environmental parameters (chemicals, nutrients, temperature, carbon dioxide, etc.) to ensure the normal function of the BNN processing unit. This can include periodically checking whether the SU input signal training set still produces the expected RU output signal. The BNN health control software 700 can be managed by the system's user administrator through the administrator interface 740.

[0086] The real-time function control software 701 is also responsible for managing the functionality of the BNN as a processing unit via the BNN function interface 730. The BNN interface 730 can be used to process the input and output of the BNN in real time, which are interfaced using the SU / RU system described previously. The BNN interface 730 can, for example, apply any necessary stimulus signal pre-processing tasks (610) and / or readout signal post-processing tasks (630).

[0087] The end user can interact with the real-time control software 701 through the user interface 741. The user interface 741 can be the same or different for the end user and / or the administrator user, but if they are the same, the administrator user can access more functions. Parameters for the real-time function control software 701 and for the BNN health control software and / or the BNN function interface can be stored in the database 720.

[0088] For advanced cognitive BNN processing unit functions that may require very low latency and high data throughput bandwidth, the BNN function interface 730 can apply additional data pre-processing 610 and / or data post-processing algorithms 630 under the control of the real-time function control software 701. In a possible embodiment, the real-time control software is not hosted on the Internet but rather operates locally. Therefore, the function control software 701 can be uploaded by an administrator using the administrator interface 740 or directly by an end user using the user interface 741.

[0089] The BNN controller computer system (also referred to herein as a "system" or automated system) can be programmed or otherwise configured to implement different BNN processing methods, such as receiving and / or combining stimulation input signals, processing these stimulation input signals, and generating and / or combining readout output signals according to a given application.

[0090] The BNN controller may be a computer system or part of a computer system that includes a central processing unit (CPU here means "processor" or "computer processor"), memory such as RAM and storage units such as hard disks, and communication interfaces for communicating with other computer systems over a communication network (e.g., the Internet or a local area network). Examples of computing systems, environments, and / or configurations include, but are not limited to, personal computer systems, server computer systems, thin clients, thick clients, handheld or laptop devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputer systems, mainframe computer systems, and the like. In some embodiments, the computer system may include one or more computer servers that operate in conjunction with many other general-purpose or special-purpose computing systems and may implement distributed computing such as cloud computing, for example, in a BNN data field. In some embodiments, the BNN controller may be integrated into a massively parallel system.

[0091] The BNN controller system is applicable to the general context of computer system executable instructions (e.g., program modules) executed by a computer system. In general, program modules can include routines, programs, objects, components, logic, data structures, etc. that perform specific tasks or implement specific abstract data types. As is well known to those skilled in the art of computer programming, program modules can use native operating system and / or file system functions, stand-alone applications; browser or application plug-ins, applets, etc.; commercial or open source libraries and / or library tools that can be programmed in Python, Biopython, C / C++ or other programming languages; custom scripts such as Perl or Bioperl scripts; and highly specialized languages ​​suitable for linear genetic programming and cognitive computing, such as SlashA or machine code or any other data structure that describes computational steps (such as those used for forward genetic programming, Cartesian genetic programming, or tree-based genetic programming).

[0092] Instructions may be executed in distributed cloud computing environments where tasks are performed by remote processing devices that are linked through a communications network. In a distributed cloud computing environment, program modules may be located in both local and remote computer system storage media including memory storage devices.

[0093] Thus, the entire system can operate as a biological operating system 750 because it integrates all the systems required to ensure the operation of the BNN computing system.

[0094] In possible embodiments, the BNN functional interface 730 may include various input and / or output signal processing algorithms, such as pre-processing or post-processing filters, classifiers, and machine learning algorithms based on mathematical or statistical models, and the functional control software 701 may control these algorithms and their parameters according to the needs of the end user. Signal pre-processing may include transforming the signals to be learned by the BNN so that they are optimally suited to the SU stimulation capabilities and format. Signal post-processing may include converting the signals output from the BNN into a more comprehensive format so that they provide the target function. In possible embodiments, the BNN may be modeled as a nonlinear system, and signal pre-processing may include applying nonlinear gains to the input signals. In possible embodiments, the BNN functional interface may include one or more artificial neural networks as pre-processors and / or post-processors, and parameters may include weight values, selection of activation functions, and other parameters that enable learning and / or classification of the signal. For example, a pre-processor may be used to repeat input application signals over a period of time or slightly alter these input application signals for more robust training and re-enhancement of the learning process.

[0095] In general, automation aims to produce a BNN processing system that generates a correct set of spatio-temporal output signals Oi(t) for a given spatio-temporal input signal Sj(t) (i identifies the stimulation input electrode, j identifies the sensing output electrode, and t identifies the sampling time). Typically, a BNN can be trained with a set of k different functions {O(t), S(t)}(k), which is collectively referred to as the training set, where O(t) and S(t) represent vectors of components Oi(t) and Si(t), respectively. After successful training, the BNN can be used to predict the correct output for inputs S(t) that are not part of the training set. Successful training, and therefore successful prediction, can be measured as minimizing the difference between the target signal O(t) and the measured BNN output signal. Various metrics can be used for this purpose, such as, for example, the mean squared error (MSE) or more generally an n-norm distance, such as the Euclidean distance or the l-norm distance. Formally, this means being able to output the correct value for a set of p different functions {O(t), S(t)}(p), which are collectively referred to as the test set that is not part of the training set.

[0096] In some embodiments, to achieve such successful training, machine learning algorithms (artificial neural networks ANN, convolutional neural networks CNN, support vector machines SVM and deep learning, but this also includes any ML methods such as random forests, genetic algorithms, genetic programming, reservoir computing, etc.) can be used to optimally present inputs and optimally measure outputs. For example:

[0097] - some of the k functions {O(t), S(t)}(k) may need to be repeated more frequently than others;

[0098] -Some inputs can be mapped to different electrodes over time;

[0099] - It may be necessary to read several electrodes simultaneously;

[0100] -etc.

[0101] More generally, depending on the application requirements, the raw signal input 135 can be fed to a first pre-processing subunit 610 of the BNN interface 730, which processes the raw signal input 135 before feeding it to the BNN 120 with the SU 110, while the BNN 120 output signal 635 read out from the RU 130 can itself be further transformed by a second post-processing subunit 630 of the BNN interface 730 to produce a more suitable output signal 135. In an open-loop architecture, the pre-processing subunit and the post-processing subunit can operate independently of each other. In a closed-loop architecture, the pre-processing subunit and the post-processing subunit can operate jointly, for example, some output from the post-processing unit can be fed back into the stimulus pre-processing unit.

[0102] Preferably, the pre-processing unit can perform spatio-temporal processing of the original input Sr(t) to generate a processed signal S(t) to be fed into the BNN with SU. The BNN then outputs the original signal O r (t), the original signal O r (t) is in turn processed by a post-processing unit to produce a processed output signal O(t). The dimensionality of each of these signals can be different. For example, in order for the BNN to output a spike for a simple, raw, one-dimensional sinusoidal signal with a frequency of 1 Hz, it may be necessary to stimulate 100 different electrodes one after another, each with a different time signal defined at a resolution of 100 Hz. The pre-processing unit 610 can be adjusted accordingly. The pre-processing unit 610 and the post-processing unit 630 can also use machine learning. In the latter embodiment, the three subsystems can be trained to achieve the overall goal of minimizing the difference between the target signal of the predefined input signal 105 and the measured final output signal 135. Note that as BNN technology develops, it may also be possible to implement the pre-processing unit 610 and / or the post-processing unit 630 in wetware rather than in software or hardware, for example as a simple BNN, in order to gradually build more complex systems suitable for learning higher cognitive functions.

[0103] Therefore, the BOS 750 integrates health monitoring and functional control of the BNN culture 120 to provide real-time sustainable and reliable BNN computing operations (including learning) to end users by implementing the following functions:

[0104] Steady-state functionality, which works by optimizing the BNN functional interface 730 and ECCU parameters 710 to ensure a constant performance level;

[0105] A learning function that works by optimizing the BNN interface (possibly including input-side signal pre-processing and / or output-side signal post-processing algorithms) along with the ECCU parameters required to learn new input / output associations. In particular, the function control software 701 is capable of adjusting the supply of additives for BNN health control and / or the values ​​of environmental parameters in real time, in closed-loop coordination with the stimulus and readout signals processed by the BNN function interface, to promote optimal learning with respect to the training set.

[0106] To maintain functionality, this may require cloning aged BNN cultures so that these BNN cultures operate directly on a training set that has already been learned by a pre-existing BNN. Then, at least a portion of the BNN interface and the ECCU parameters of the pre-existing BNN can be retrieved from the database 720 and copied to accelerate learning time on the new BNN.

[0107] BNN automated learning

[0108] The inherent dynamic interconnectedness of neuronal cells is viewed as a means to create (re)programmable functions based on data feedback, which is an essential component of machine learning. It will be apparent to those skilled in the art of neuroscience that a BNN system can be first trained with input data signals 105 until the BNN system reaches a stable state that gives a desired output data signal 135, even for inputs not fed during training. When the BNN is able to relate the target output data signal to the input data signal as desired, the system is said to have learned and generalized because it can produce a deterministic response when fed a given input.

[0109] To train the BNN, the automated controller may adapt the input spatiotemporal signal 605 of the stimulation unit SU to the input data signal 105 and the output spatiotemporal data signal 135 until the output spatiotemporal data signal 135 matches the target output data signal. The automated controller may further adjust at least one or more of the following:

[0110] -BNN core unit environmental parameters;

[0111] - Nutrition supply for BNN neural cell culture;

[0112] -BNN neural cell culture supplement supply;

[0113] -BNN neural cell culture nutrient waste collection;

[0114] -BNN neural cell culture additive waste collection;

[0115] -BNN interface signal preprocessing algorithm parameters;

[0116] -BNN interface signal post-processing algorithm parameters;

[0117] Until the output spatiotemporal data signal matches the expected output data signal.

[0118] Figure 8 A learning process that can be implemented by real-time processing software is shown, where a feedback loop H 800 adapts the stimulation signal 605 to the measured readout signal 635. In a possible embodiment, the real-time processing software can trigger periodic repetition of the stimulation signal 605 on the BNN to promote long-term potentiation. More generally, environmental parameters and chemical parameters controlled by the health monitoring system can also be part of the closed-loop learning system. In a possible embodiment, when the measured readout signal 635 is consistent with the training set expected signal for a given stimulation signal input, the real-time processing software can trigger the delivery of a drug dose known to enhance the BNN in return.

[0119] In other possible embodiments (not shown), the BNN can be connected to an external information source, such as an Internet web database, via a BNN automation controller interface. The BNN can further learn to access and use this external information as a source to learn more, connect new concepts, and further develop high-level cognitive abilities over time.

[0120] BNN Biocomputing Stack

[0121] In order for the feedback loop to work, closed-loop control of the BNN core unit is necessary. The proposed automated vascularization system may be limited to the growth and maintenance of relatively small BNN cultures compared to the mammalian brain, for example, on the order of 10,000 interconnected neurons. Therefore, it may not be sufficient to integrate only one BNN core unit into a BNN computing system to learn and operate high-level cognitive functions. This limitation can be overcome by a BNN biocomputing stack (BCS), such as Figure 9a )and Figure 9b ) is shown. In a possible embodiment, one or more BNN core units can be arranged in series. The readout unit RU of the first BNN core unit can be connected to the stimulation unit SU of the next BNN core unit, or alternatively, they can be combined in a single hardware. At each interface between blocks, the system can also accept Figure 9b ) as shown in the external stimulation signal input 900 (ES).

[0122] As a possible embodiment, 2D BNN core units can be stacked vertically. In another possible embodiment, BNN core units fabricated as neurosphere stacks or 3D bioprinted material layers can be mechanically arranged as a series stack, with electrophysiological probes inserted at the interface between two adjacent units.

[0123] like Figure 9b ), in an arrangement similar to a deep learning layered architecture, the learning process can be controlled end-to-end via a feedback loop operating between the final readout unit and the first stimulation unit. In this configuration, due to the inherent properties of biological neural networks, the feedback element acts as a stabilizer, changing its internal state even in the absence of external stimulation.

[0124] Figure 10a )and Figure 10b ) shows two additional embodiments of an adaptive closed-loop system for controlling a BNN computation stack with additional nonlinear gain P 1000, with and without an additional post-processing block O 1010, respectively. As will be apparent to those skilled in the art of deep learning, the pre-processing block P can facilitate the learning process, and the post-processing block O can represent the topology for reservoir computation. This configuration supports a trainable biological computation stack (T-BCS).

[0125] Figure 11 Shown with Figure 10b ) related specific implementation, wherein the O output signal post-processing block 1010 is implemented by an artificial neural network (ANN). The biological neural network (BNN) 120 is connected to the digital interface readout unit (RU) and the stimulation unit (SU) through a multi-electrode array (MEA) 1135. The feedback function (HI) 1137 corresponds to the learning process of the ANN, such as but not limited to backpropagation. The external feedback function (H2) 1136 corresponds to the average value of the long-term potential applied to the BNN. Genetic programming or machine learning can be used to implement the external feedback function H2 in order to determine the correct spatiotemporal sequence of desired spiking at the output of the BNN culture 120. Another way to apply long-term potentiation is to use RU as a stimulation unit with a desired spike sequence, which will strengthen the internal connections inside the BNN. Using this approach, the ANN can even be omitted in some embodiments.

[0126] Let S(t) be the number of different stimulation units S that depend on MEA i Let O(t) be the periodic data input function of time (t) for a given S(t) for different readout units O of MEA. i Let us say that the data output function depends on time (t).

[0127] Then, H2 is a function or algorithm (found by example with the help of genetic programming or machine learning) that minimizes (or maximizes) a scalar metric L, such that the BNN always produces the same spike time function O(t) for a given input S(t) when L is minimized (or maximized) for any number of periods P. L is called the loss (or reward) function.

[0128] For example, L could be:

[0129]

[0130] Where N is the period of the input function S(t). In this case, the function or algorithm H2 will minimize L.

[0131] Figure 12 Shown Figure 11Another possible embodiment of the implementation is to facilitate long-term maintenance of T-BCS operation. In practice, the BNN culture 120 may evolve over time and deviate from its initial (learned) structure, thus potentially causing drift in the T-BCS data output. This can be monitored by periodically testing the T-BCS operation using a validation set that includes one or more of the initial signal training set as SU input. The output of the T-BCS readout unit RU can then be checked against the expected prediction set, and if the difference is too large, the T-BCS must reapply the learning process. Since the learning process takes time and disrupts T-BCS functionality, in a possible embodiment, the T-BCS system is replaced with a new system that has been prepared in advance.

[0132] BNN Server

[0133] The proposed automated BNN system, BNN operating system (BOS), and trainable BNN computing stack (T-BCS) can provide the core architecture for network-based computing servers. Compared with traditional software- and hardware-based server architectures, such BNN servers may be better suited to provide advanced cognitive processing more efficiently.

[0134] like Figure 13 As shown, the T-BCS wetware architecture can operate with a host that provides different services to user clients. Figure 14 Possible functions that can be operated by the host server to manage user clients are also described, such as processing requests, scheduling jobs, reporting and logging, maintaining dashboards, and invoicing. Figure 15 A general architecture with load balancing is also described for serving multiple clients from the same T-BCS server host.

[0135] Preferably, the BNN server supports redundant T-BCS operation, such that one T-BCS can be removed for maintenance (eg, relearning) while at least one other T-BCS remains operational to service client requests.

[0136] BNN maintenance and update example applications

[0137] Figure 16 b) shows a possible application of the T-BCS server as a reverse image search service compared to the traditional deep learning server 16a). Figure 16 a) Traditionally, a reverse image search service, such as that implemented by Google, takes an image as input and tells you where that image is located on the web, gives a description of the image, and may also return a collection of similar images. Figure 16The prior art reverse image search backend of a) uses multiple different image processing or image vision systems 1613 in order to obtain a robust and compact representation of the image, typically as a hash value 1614. The proposed application replaces the complete backend processing with T-BCS 1615, significantly reducing power consumption at the same computing power. In particular, once properly trained, the Bio-Computing Stack (BCS) can advantageously replace the feature extraction part, including the machine learning AI network, or the feature descriptor extractor algorithm, or the perceptual hash function, or additional image processing tasks, or any combination of these tasks, as well as the so-called hashing post-processing required to obtain a compact representation of the image to be stored in the database.

[0138] Other experiments, examples and applications

[0139] Figure 17 Rat cortical stem cells were shown to form neurospheres approximately 200 meters wide after maturation for 4 days in DMEM-F12 medium supplemented with glutamine (TM), fibroblast growth factor, epidermal growth factor, and Stempro (R). In a possible embodiment, 3D electrodes can be distributed on the surface of neurospheres (or any other type of neuronal aggregates). After growth stimulation, the tips of the electrodes will gradually fit within the growing neurospheres. Figure 18 、 Figure 19 and Figure 20 Schematic showing an example of 12 electrodes regularly distributed along a virtual ring of a neurosphere of cortical neural stem cells that grew from 400 μm to over 1 mm in 15 days, allowing the tip of each electrode to naturally become more deeply embedded in the neurosphere. In this particular case, growth is driven by stromal-stimulated, such as stroma extracted from Engelbreth-Holm-Swarm (EHS) mouse sarcoma.

[0140] It is also possible to use matrices to stimulate 3D growth of neural cells directly on the MEA. Figure 21 , wherein the adherent cells 2120 grow, such that the adherent cells 2120 extend their axons 2121, 2122, or the adherent cells 2120 migrate through Matrix 2100. This enables much thicker networks to be obtained, potentially extending more than a few millimeters above the MEA surface 210. Figure 22 Adherent cells 2201 and Microscope images of a 3D image ...

[0141] Although various embodiments have been described above, it should be understood that they are presented by way of example and not limitation. It will be apparent to those skilled in the relevant art that various changes in form and detail may be made without departing from the spirit and scope. Indeed, after reading the above description, those skilled in the relevant art will understand how to implement alternative embodiments.

[0142] It is obvious to those skilled in the art of digital data communication that the methods described herein can be applied indiscriminately to various data structures such as data files or data streams. Therefore, the terms "data", "data structure", "data field", "file" or "stream" can be used indiscriminately in this specification.

[0143] Although the above detailed description contains many specific details, these should not be construed as limiting the scope of the embodiments but rather as merely providing illustrations of some of several embodiments.

[0144] Although various embodiments have been described above, it should be understood that they are presented by way of example and not limitation. It will be apparent to those skilled in the relevant art that various changes in form and detail may be made without departing from the spirit and scope. Indeed, after reading the above description, those skilled in the relevant art will understand how to implement alternative embodiments.

[0145] Furthermore, it should be understood that any drawings highlighting features and advantages are presented for illustrative purposes only. The disclosed methods are sufficiently flexible and configurable that they can be used in ways other than those shown.

[0146] Although the term "at least one" is often used in the specification, claims, and drawings, the terms "a," "an," "the," "said," etc. also mean "at least one" or "the at least one" in the specification, claims, and drawings.

[0147] Throughout this specification, multiple instances may implement components, operations, or structures described as a single instance. Although the individual operations of one or more methods are described and described as separate operations, one or more of the individual operations may be performed simultaneously and need not be performed in the order described. The structures and functions presented as separate components in the example configurations may be implemented as combined structures or components. Similarly, the structures and functions presented as single components may be implemented as separate components. These and other variations, modifications, additions, and improvements fall within the scope of this paper's subject matter.

[0148] Certain embodiments are described herein as including logic or a number of components, modules, units, or mechanisms. A module or unit may constitute a software module (e.g., code embodied on a machine-readable medium or in a transmission signal) or a hardware module. A hardware module is a tangible unit capable of performing a specific operation and may be configured or arranged in a specific physical manner. In various example embodiments, one or more computer systems (e.g., a stand-alone computer system, a client computer system, or a server computer system) or one or more hardware modules of a computer system (e.g., a processor or a group of processors) may be configured by software (e.g., an application or application portion) as a hardware module that operates to perform certain operations described herein.

[0149] In some embodiments, the hardware module can be implemented mechanically, electronically, biologically, or any suitable combination thereof. For example, the hardware module may include a dedicated circuit or logic that is permanently configured to perform certain operations. For example, the hardware module may be a dedicated processor, such as a field programmable gate array (FPGA) or an ASIC. The hardware module may also include programmable logic or circuits that are temporarily configured by software to perform certain operations. For example, the hardware module may include software contained in a general-purpose processor or other programmable processor. It will be understood that the decision to implement the hardware module mechanically, in a dedicated and permanently configured circuit, or in a temporarily configured circuit (e.g., configured by software) may be driven by cost and time considerations. With the development of synthetic biology, all or part of the hardware module may be made from biological cells, such as neurospheres and / or genetically engineered cells (also referred to as wetware).

[0150] The various operations of the example methods described herein may be performed, at least in part, by one or more processors that are temporarily configured (e.g., by software) or permanently configured to perform the relevant operations. Regardless of whether such processors are temporarily or permanently configured, such processors may constitute processor-implemented modules that operate to perform one or more operations or functions described herein. As used herein, a processor-implemented module refers to a hardware module that is implemented using one or more processors.

[0151] Similarly, the methods described herein may be at least partially implemented by a processor, which is an example of hardware. For example, at least some operations of the methods may be performed by one or more processors or processor-implemented modules.

[0152] Some portions of the subject matter discussed herein may be presented in terms of algorithms or symbolic representations of operations on data stored as bits or binary digital signals within a machine memory (e.g., a computer memory). Such algorithms or symbolic representations are examples of techniques used by those of ordinary skill in the data processing arts to convey the substance of their work to others skilled in the art. As used herein, an algorithm is a self-consistent sequence of operations or similar processes leading to a desired result. In this context, algorithms and operations involve physical manipulation of physical quantities.

[0153] Although an overview of the inventive subject matter has been described with reference to specific example embodiments, various modifications and variations may be made to these embodiments without departing from the broader spirit and scope of the embodiments of the present invention. For example, one of ordinary skill in the art may mix and match the various embodiments or features thereof, or make them optional. These embodiments of the inventive subject matter may be referred to herein, individually or collectively, by the term "invention," which is merely a matter of convenience and is not intended to limit the scope of this application to any single invention or inventive concept, if more than one is actually disclosed.

[0154] It is believed that the embodiments described herein are described in sufficient detail to enable those skilled in the art to practice the disclosed teachings. Other embodiments may be used and derived therefrom, such that structural and logical substitutions and changes may be made without departing from the scope of the present invention. Therefore, the detailed description should not be construed as limiting, and the scope of the various embodiments is limited only by the appended claims and the full scope of equivalents to which such claims are entitled.

[0155] In addition, multiple instances may be provided for resources, operations, or structures described herein as single instances. In addition, the boundaries between various resources, operations, modules, engines, and data stores are somewhat arbitrary, and specific operations are shown in the context of specific illustrative configurations. Other functional allocations are foreseeable and may fall within the scope of various embodiments of the present invention. In general, structures and functions presented as separate resources in the example configurations may be implemented as combined structures or resources. Similarly, structures and functions presented as single resources may be implemented as separate resources. These and other variations, modifications, additions, and improvements fall within the scope of the embodiments of the present invention as represented by the appended claims. Therefore, the description and drawings are to be considered illustrative rather than restrictive.

[0156] Finally, it is Applicant's intent that only claims containing the express language "means for" or "step for" be construed under 35 U.S.C. § 112, paragraph 6. Claims that do not expressly include the phrase "means for" or "step for" shall not be construed under 35 U.S.C. § 112, paragraph 6.

Claims

1. An automated processing system for transforming a spatiotemporal input data signal (105) into a spatiotemporal output data signal (135), the system comprising: - an in vitro biological neural network BNN (100) culture of neural cells in a BNN core unit (120); - an input stimulation unit SU (110) adapted to apply an input spatiotemporal stimulation signal (605) to the first group of nerve cells; - an output readout unit RU (130) adapted to capture the output spatiotemporal readout signal (635) from the second group of neural cells; - one or more nutrient tanks (319) connected to one or more nutrient dispensers (320) for injecting one or more nutrients into the biological neural cell culture (120); - one or more additive tanks (321, 323), each of which is connected to one or more additive dispensers (322, 324) for injecting one or more additives into the BNN culture (120); - one or more nutrient waste collectors (325) for filtering and draining nutrient waste from the BNN culture (120); - one or more additive waste collectors (326, 327) for filtering and draining additive waste from the BNN culture (120); - one or more vascularized networks for connecting the nutrient distributor (320), additive distributors (322, 324), nutrient waste collector (325) and additive waste collector (326, 327) to the BNN culture; - one or more sensors for measuring at least one environmental parameter of said BNN culture (120); - an automated controller (600) configured to adapt the stimulation signal (605) to the input data signal (105), to adapt the output data signal (135) to the readout signal (635), and to control at least one of: -BNN core unit environmental parameters; - Nutritional supply of BNN culture; -BNN culture supplement supply; -BNN culture nutrient waste collection; -BNN culture additive waste collection; to maintain homeostasis of the BNN culture over time so as to continuously convert said spatiotemporal input data signal (105) into said spatiotemporal output data signal (135), wherein the automation controller (600) further comprises a pre-processing unit that converts the input data signal (105) into the stimulation signal (605) and wherein the signal pre-processing comprises applying a time-repeating input data signal (105) or a time-varying input data signal (105), wherein one or more additive dispensers (322, 324) comprise valves, syringes, or pumps for delivering additives to the BNN culture, And wherein the automated controller (600) is configured to adjust the additive supply in real time until the spatiotemporal output data signal (135) matches a desired output data signal.

2. The automated processing system according to claim 1, wherein: The nutrients are selected from amino acids, carbohydrates, vitamins and minerals or combinations thereof.

3. The automated processing system according to claim 1, wherein: The one or more additives are selected from dopaminergic stimulation enhancers to increase the dopaminergic response of the BNN, dopaminergic stimulation inhibitors to reduce the BNN, a dose of a drug known to enhance the BNN in return when the spatiotemporal output data signal (135) matches the expected output data signal for a given stimulation signal input, botulinum toxin, nicotine, curry, amphetamines, cocaine, MDMA, strychnine, THC, caffeine, benzodiazepines, barbiturates, alcohol, opioids, growth factors, hormones, gases, or a combination thereof.

4. The automated processing system according to claim 1, wherein: The vascularized networks were fabricated using 3D bioprinting with biocompatible materials.

5. The automated processing system according to claim 1, wherein: The vascularized network grows from stem cells on a BNN (120) culture support.

6. The automated processing system according to any one of claims 1 to 3, wherein: The vascularized network is a soft and compressible, porous and absorbent material that forms a 3D structure suitable for mechanically accommodating BNN cells in its pores while regulating the growth and development of the BNN cells as a 3D cell culture.

7. The automated processing system according to any one of claims 1 to 3, wherein: Sensors measure temperature, humidity, pH or CO2 environmental parameters of the BNN culture.

8. The automated processing system according to any one of claims 1 to 3, wherein: The preprocessing unit transforms the input data signal (105) into the stimulation signal (605) using at least one of a spatiotemporal signal filter, a spatiotemporal signal classifier, a machine learning algorithm based on a mathematical or statistical model, an artificial neural network, a convolutional neural network, a support vector machine classifier, a random forest classifier, a genetic algorithm, a genetic programming algorithm or a reservoir calculation method.

9. The automated processing system according to any one of claims 1 to 3, wherein: The automated controller further comprises a post-processing unit that transforms the readout signal (635) into the output data signal (135) using at least one of a spatiotemporal signal filter, a spatiotemporal signal classifier, a machine learning algorithm based on a mathematical or statistical model, an artificial neural network, a convolutional neural network, a support vector machine classifier, a random forest classifier, a genetic algorithm, a genetic programming algorithm, or a reservoir computing method.

10. An automated processing system according to any one of claims 1 to 3, wherein the automated processing system is further suitable for controlling the stacking of BNN core units, the first part of the BNN culture is in a first BNN core unit from the stack, and the second part of the BNN culture is in a second BNN core unit from the stack.

11. A server for executing automated processing tasks, the server comprising the automated processing system according to any one of claims 1 to 10.

12. A method for converting a spatiotemporal input data signal (105) into a spatiotemporal output data signal (135) using an automated controller (600) and a biological neural network (BNN) core unit (120), the BNN core unit comprising at least an in vitro culture of neural cells, i.e., a BNN culture, the in vitro culture of neural cells being adapted to feed a stimulation spatiotemporal signal (605) into a first group of neural cells using an input stimulation unit SU (110) and to read out a spatiotemporal signal from a second group of neural cells using an output readout unit RU (130), the method comprising: - pre-processing the spatiotemporal input data signal (105) into the stimulus spatiotemporal signal (605) using the automated controller (600); and wherein the pre-processing comprises applying a time-repeating input data signal (105) or a time-varying input data signal (105), - applying a stimulation spatiotemporal signal (605) to a first group of neural cells of an in vitro culture of neural cells, i.e., a BNN culture (120), using an input stimulation unit SU (110); - reading the spatiotemporal signal (635) from the second group of neural cells using the output readout unit RU (130); - post-processing the read-out spatiotemporal signal (635) into the spatiotemporal output data signal (135) using the automation controller (600); - using the automation controller (600) to control at least one of the following: BNN core unit environmental parameters; Nutritional supply for BNN cultures; BNN culture supplement supply; BNN culture nutrient waste collection; BNN culture additive waste collection; Preprocessing parameters; Post-processing parameters; To maintain the homeostasis of the BNN neural cell culture over time, so that the BNN trusted unit continuously converts the spatiotemporal input data signal (105) into the spatiotemporal output data signal (135); - injecting one or more additives from one or more respectively connected additive tanks (321, 323) into the BNN culture (120) using one or more additive dispensers (322, 324); -Receive target spatiotemporal output data signal; - using the automated controller (600) to adjust the additive supply in real time until the spatiotemporal output data signal (135) matches the target spatiotemporal output data signal.

13. The method according to claim 12, further comprising: -Receive target spatiotemporal output data signal; -Adjust at least one of the following: BNN core unit environmental parameters; Nutritional supply for BNN cultures; BNN culture supplement supply; BNN culture nutrient waste collection; BNN culture additive waste collection; Preprocessing parameters; Post-processing parameters; to minimize the error between the spatiotemporal output data signal (135) and the target spatiotemporal output data signal, and The one or more additives include a dose of a drug known to enhance the BNN in return when the spatiotemporal output data signal (135) matches an expected output data signal for a given stimulation signal input.

14. The method according to claim 12, wherein: The preprocessing includes transforming the input data signal (105) into the stimulation spatiotemporal signal (605) using at least one of a spatiotemporal signal filter, a spatiotemporal signal classifier, a machine learning algorithm based on a mathematical or statistical model, an artificial neural network, a convolutional neural network, a support vector machine classifier, a random forest classifier, a genetic algorithm, a genetic programming algorithm, or a reservoir computing method.

15. The method according to any one of claims 12 to 14, wherein The post-processing includes transforming the read-out spatiotemporal signal (635) into the output data signal (135) using at least one of a spatiotemporal signal filter, a spatiotemporal signal classifier, a machine learning algorithm based on a mathematical or statistical model, an artificial neural network, a convolutional neural network, a support vector machine classifier, a random forest classifier, a genetic algorithm, a genetic programming algorithm, or a reservoir computing method.

Images