1794 results about "Artificial neuronal network" patented technology

The invention provides a composite material performance prediction and process optimization method based on a neural network, and the method comprises the steps: firstly collecting multi-source data in the preparation and test process of a composite material, carrying out the preprocessing of the data, screening key feature variables as input variables, constructing a feedforward artificial neural network model, and predicting and outputting the performance indexes of the composite material. And training the model, performing iterative optimization on model parameters, and optimizing composite material process parameters by using the optimized model based on a reverse optimization strategy of a genetic algorithm to obtain an optimal process parameter combination. The invention provides a scientific, efficient and reliable tool for design and optimization of composite materials, and particularly has wide application prospects in high-requirement industries such as aerospace and the like.

A system for an automated real-time fraud detection based on predictive analytics of call data including a processor of a call analysis server (CAS) node configured to host a machine learning (ML) module and connected to at least one user-entity node over a network and a memory on which are stored machine-readable instructions that when executed by the processor, cause the processor to: capture user call data including a plurality of key elements from the at least one user-entity node; parse the user call data to extract a plurality of classifying features based on the plurality of key elements; query a local calls database to retrieve local historical calls-related data based on the plurality of classifying features; generate at least one feature vector based on the plurality of classifying features and the local historical calls-related data; provide the at least one feature vector to the ML module coupled to an Artificial Neural Network (ANN); receive a plurality of user call ranking parameters from a call predictive model generated by the ML module using outputs of the ANN based on the feature vector; and generate a fraud verdict for the user call based on the plurality of call ranking parameters.

Aspects of this disclosure relate to digital compensators, such as digital predistortion systems. Digital predistortion systems disclosed herein use a neural Volterra approach. Such digital predistortion systems can include a feature processing path comprising a feature artificial neural network, an envelope processing path, multipliers configured to multiply respective output signals of the feature processing path and the envelope processing path, and a combiner configured to generate a combined output signal based on at least output signals of the multipliers. The combined output signal is a digitally predistorted version of an input signal.

A hardware emulator and an emulation system including the hardware emulator are provided. The hardware emulator includes an artificial neural network-based reconstruction model configured to reconstruct dynamics of a dynamical system based on input data and a memristor-based circuit configured to emulate state space representation of the dynamical system based on the reconstruction model.

A method for quantifying lung function using a medical image, comprising the following steps: capturing or receiving a medical image that includes anatomical information for a lung region of a patient; segmenting at least one region of abnormality in the lung region of the medical image using an artificial neural network; and predicting a quantification result with respect to lung function based on the size of the at least one region of abnormality.

An inspection method for a lithium secondary battery can improve the reliability, accuracy, and reproducibility of inspection results by advancing a learning of an artificial neural network according to one or more divided regions of one or more surfaces of the lithium secondary battery and one or more types of defects occurring in each region.

A set of values can be selected from a plurality of fields of a table in a database. At least one adverb or adjective can be selected for the set of values. Join paths for values in the set of values can be determined. A structured query language pattern can be determined based, at least in part, on at least one value in the set of values, the at least one adverb or adjective for the set of values, and the join paths for the set of values. The structured query language pattern can be stored to first training data configured, at least in part, for use in machine learning to train a text-to-SQL model, the text-to-SQL model comprising a first artificial neural network and configured to convert first natural language text to a first structured query language query.

The invention is suitable for the technical field of thunderstorm prediction, and provides a strong thunderstorm potential risk and intensity prediction method based on an artificial neural network, and the method comprises the steps: obtaining the ground-to-ground lightning data of a target region, determining the time distribution characteristics of a thunderstorm event, and recognizing the time valley of the thunderstorm event; collecting multi-source meteorological data in a target area, and screening meteorological prediction factor data; performing space-time alignment processing on the ground-to-ground lightning data and the meteorological prediction factor data, and constructing a grid unit day-by-day sample set based on the aligned data; constructing a probability classification model of a strong thunderstorm event based on the grid unit day-by-day sample set; building a regression neural network model based on the grid units which are judged to be strong thunderstorm high risks by a probability classification model; and a collaborative prediction result is output, a predicted value of a geometric mean value of the ground-to-ground flash frequency and the lightning current amplitude is synchronously output, and a risk assessment basis of classification discrimination and numerical prediction capability is provided for extreme weather early warning and power grid disaster prevention and reduction through double-order modeling.

The invention provides a reservoir pressure salinity scheduling method based on a physical information neural network (PINN). The reservoir pressure salinity scheduling method comprises the following steps: Step 1, salinity prediction based on a physical information neural network (PINN) model; the method comprises the following steps: in a PINN model framework, embedding a physical law of salinity conservation into a multi-layer perceptron (MLP) artificial neural network for training so as to carry out salinity prediction; step 2, establishing an estuary water-salt model based on a three-dimensional ocean numerical FVCOM model; step 3, carrying out upstream reservoir salty water pressing and light water supplementing emergency scheduling based on salinity forecast; comprising the following steps: Step3.1, establishing a multi-objective function; and Step3.2, solving a scheduling model. According to the method, the salinity momentum conservation constraint can be considered, the salinity prediction accuracy under the influence of different upstream flows and downstream tidal ranges can be improved, and the emergency scheduling of the reservoir for the salt tide invasion of the estuary drinking water source can be carried out in combination with the scheduling model, so that the water supply safety is ensured.

A novel and useful system and method of improved power performance and lowered memory requirements for an artificial neural network based on packing memory utilizing several structured sparsity mechanisms. The invention applies to neural network (NN) processing engines adapted to implement mechanisms to search for structured sparsity in weights and activations, resulting in a considerably reduced memory usage. The sparsity guided training mechanism synthesizes and generates structured sparsity weights. A compiler mechanism within a software development kit (SDK), manipulates structured weight domain sparsity to generate a sparse set of static weights for the NN. The structured sparsity static weights are loaded into the NN after compilation and utilized by both the structured weight domain sparsity mechanism and the structured activation domain sparsity mechanism. The application of structured sparsity lowers the span of search options and creates a relatively loose coupling between the data and control planes.

Disclosed is a method of processing a query using a trained artificial neural network (ANN) implementing a large language model (LLM). A database of documents is maintained for use in processing queries, the database defining multiple repositories, each containing one or more documents. The documents in the database are processed to generate, for each repository, a vector store comprising vector embeddings encoding information obtained from documents of the repository. A user inputs a query string and a selection of one or more of the repositories to be used to process the query. A query embedding corresponding to the query string is generated and the vector stores corresponding to each selected repository are searched using the query embedding to identify one or more vectors that are relevant to the query. An LLM query is formulated to include the query string, a query context comprising information determined based on the identified relevant vectors and a predefined LLM prompt. The LLM is invoked with the LLM query as input whereby the LLM query is processed using the ANN to generate an LLM output. A query response is provided to the user based on query response data received from the LLM.

A processor-implemented method includes receiving an artificial neural network having a number of pre-trained weights. The method also includes training a subset of the number of pre-trained weights to obtain trained sparse adapter weights for obtaining a fine-tuned version of the artificial neural network. The subset of the number of pre-trained weights includes base model weights of a base model for the artificial neural network. The subset of the number of pre-trained weights is selected with a sparse mask of a sparse adapter. The method may also include replacing the subset of the number of pre-trained weights with the trained sparse adapter weights.

A regional building cooling and heating load artificial neural network energy distribution optimization method disclosed by the present invention comprises the steps of obtaining inter-regional topological relation data and load-related data, constructing a graph network model to capture spatial dependency, fusing meteorological data and historical load records to generate a multivariable dynamic data set, and obtaining the optimal energy distribution of the cooling and heating load artificial neural network. Processing the multivariable dynamic data set by adopting a graph convolutional neural network to obtain an initial prediction result, adjusting the initial prediction result through an artificial neural network to generate adjusted load distribution data, and analyzing by combining a heat transfer rule and a time sequence to obtain spatial and temporal distribution characteristics; and integrating equipment efficiency constraint and cost fluctuation data by adopting a multi-objective optimization algorithm to obtain an optimized objective function value, adjusting equipment dynamic parameters through deep reinforcement learning to obtain a parameter adjustment scheme, and optimizing an energy distribution proportion by adopting a linear programming algorithm to obtain a final energy distribution scheme.

The invention discloses an intelligent rehabilitation training method, system and device based on a wearable hemiplegic patient, and relates to the technical field of rehabilitation training, and the method comprises the steps: collecting multi-dimensional data of the patient, carrying out the preprocessing, obtaining an evaluation value of the multi-dimensional data through an artificial neural network model, carrying out the splicing, generating an evaluation vector, and carrying out the recognition of the evaluation vector; performing optimization as an individual of a bald eagle search optimization algorithm to obtain an optimal evaluation value vector; defining an evaluation value in the optimal evaluation value vector as an input variable of fuzzy logic to perform fuzzy reasoning, forming a setting vector and a training target vector, constructing Bayesian prior distribution, a likelihood function and Bayesian posteriori distribution to perform maximization solution, obtaining an optimal training target vector to perform linear mapping, generating a specific value vector, and obtaining the optimal training target vector. A rehabilitation training task and dynamic adjustment and feedback of a specific value vector are executed; according to the invention, effective optimization and personalized customization of the rehabilitation training process are realized, and the efficiency and effect of rehabilitation training are improved.

A method for classifying neurological disease status is described. The method includes acquiring, by a data preprocessor logic, patient image data. The method further includes generating, by a trained artificial neural network (ANN), a classification output based, at least in part, on the patient image data. The classification output corresponds to a neurological disease status of the patient. The trained ANN is trained based, at least in part, on longitudinal source data.

The invention discloses a manufacturing defect-oriented multi-scale composite material mechanical property prediction method, and belongs to the technical field of composite material mechanical property analysis. The method comprises the following steps: based on a multi-scale representative volume element (RVE) theory, respectively constructing parameterized models containing pores and fiber dislocations in a microscopic scale and a mesoscale; the equivalent elastic parameters under each scale are calculated through a finite element numerical homogenization technology, and an agent model is constructed in combination with an artificial neural network, so that the calculation cost is remarkably reduced, and the high-efficiency cross-scale prediction of the rigidity and strength performance of the composite material is realized. The method overcomes the defect of defect modeling of a traditional model, has high precision and high efficiency, and is suitable for performance evaluation and optimization design of composite material structures in the fields of aerospace, rail transit and the like.

The invention relates to a path planning method and device based on hierarchical reinforcement learning, equipment and a medium, and the method comprises the steps: converting any planar map into rasterization, taking the shortest path between a starting point position and an end point position as an optimization target under the constraint of avoiding the collision of an obstacle, and starting from the starting point position, selecting a plurality of sub-target positions, the method comprises the following steps of: establishing a path planning problem between a starting point position and an end point position until the end point position is reached, converting the path planning problem between the starting point position and the end point position into a plurality of local sub-path planning problems, constructing a learning heuristic network by adopting an artificial neural network as a heuristic function in an A * algorithm, and training the learning heuristic network by utilizing reinforcement learning to obtain a learning heuristic network; and solving each local sub-path planning problem by using the trained learning heuristic network to obtain an optimal solution of each local sub-path, and then obtaining a bounded suboptimal path planning result. By adopting the method, path planning can be efficiently carried out on any plane map.

Methods, systems, and devices for wireless communications are described. A user equipment (UE) may receive a control signal configuring a low-density pattern for channel state information (CSI) reference signal (RS) reception for a set of multiple antenna ports. The low-density pattern may indicate a subset of the set of multiple antenna ports for the CSI-RS reception via one or more resource blocks (RBs). The UE may receive a set of multiple CSI-RSs that is based on the low-density pattern. In some cases, the UE, or some other training device, may train an artificial neural network to process the CSI-RSs according to the low-density pattern. The artificial neural network may be an example of a generalized neural network or a neural network specific to one or more low-density patterns. The UE may transmit a CSI report based on processing the CSI-RSs according to the low-density pattern.

Systems and methods for developing a game of chance. The game of chance being at least partially developed using specialized artificial intelligence (AI) game design systems or specialized artificial intelligence game design system modules or components which may include different types of machine learning or training techniques, including supervised, unsupervised, reinforced, deep learning and artificial neural networks and / or similar and may include analyzing past game performance utilizing full, partial or estimated prior game performance data for developing game math models, game mechanics and associated game programming, game art and graphics, game animations, game sound effects, partial or full game development, game computer code, game quality assurance diagnosis and editing, game analytics, compliance, help screens, and predictive models, etc.

An object of initial unknown position on a map may be determined by traversing through moving and turning to establish motion trajectory to reduce its spatial uncertainty to a single location that would fit only to a certain map trajectory. An artificial neural network model learns from object motion on different map topologies may establish the object's end-to-end positioning from embedding map topologies and object motion. The proposed method includes learning potential motion patterns from the map and perform trajectory classification in the map's edge-space. Two different trajectory representations, namely angle representation and augmented angle representation (incorporates distance traversed) are considered and both a Graph Neural Network and an RNN are trained from the map for each representation to compare their performances. The results from the actual visual-inertial odometry have shown that the proposed approach is able to learn the map and localize the object based on its motion trajectories.

The invention discloses a method and a device for calculating viscous stress at a near wall surface of a Cartesian grid solver, and aims to solve the problem of high calculation cost during viscous stress calculation in the prior art. The method comprises the following steps: acquiring a body-fitted grid and a Cartesian grid, solving a wall-intersection-free unit in the Cartesian grid by adopting a Cartesian grid solver corresponding to the Cartesian grid, and determining flow variable distribution of a low-precision flow field without the wall-intersection unit; based on the body-fitted grid, performing high-precision flow field flow variable distribution solving on the wall-surface-intersection-containing units in the Cartesian grid by adopting a data driving method based on an artificial neural network or a physical information neural network method according to low-precision flow field flow variable distribution without the wall-surface-intersection-unit, and outputting target flow field flow variable distribution; and calculating the viscosity stress according to the flow variable distribution of the target flow field.

A stacked processor-plus-memory device includes a processing die with an array of processing elements of an artificial neural network. Each processing element multiplies a first operand—e.g. a weight—by a second operand to produce a partial result to a subsequent processing element. To prepare for these computations, a sequencer loads the weights into the processing elements as a sequence of operands that step through the processing elements, each operand stored in the corresponding processing element. The operands can be sequenced directly from memory to the processing elements or can be stored first in cache. The processing elements include streaming logic that disregards interruptions in the stream of operands.

The invention provides a bay water health status evaluation method, device and product, and the method comprises the following steps: S1, determining a sampling site and collecting a zooplankton eDNA sample to obtain amplicon sequence variant information; s2, annotating amplicon sequence variant information to obtain species composition information and abundance data; s3, obtaining a first evaluation candidate index according to the species composition information and the abundance data; constructing a co-occurrence network and determining a second evaluation candidate index according to the network topology parameters; combining the first evaluation candidate index with the second evaluation candidate index to obtain a third evaluation candidate index; s4, selecting a reference point and a damaged point, and obtaining key evaluation candidate indexes through an artificial neural network model; and S5, constructing a zooplankton integrity index based on the key evaluation candidate indexes, and evaluating the water quality health grade according to the zooplankton integrity index. By utilizing the technical scheme, the precision of evaluating the ecological health condition of the severely polluted bay water body can be improved, and meanwhile, the complexity of an evaluation index system is reduced.

The invention discloses an inverter power supply network construction type control method based on an artificial neural network, and relates to the technical field of inverter power supply network construction type control. Collecting an electrical parameter sequence set of an inverter power supply grid-connected point; constructing a power supply network construction control analyzer based on the deep feedforward neural network, and outputting a first network construction control parameter; building a power distribution network simulation topology model based on PSCAD to perform parameter optimization, and generating a second network construction control parameter; and analyzing and determining the fluctuation degree of the operation state, carrying out weighted fusion on the two types of control parameters according to a set dynamic fitting strategy, obtaining an adaptive networking control strategy, and implementing regulation and control. According to the invention, on-line intelligent optimization of control parameters is realized through a two-way parallel mechanism of quick response of the neural network and accurate verification of the simulation model in combination with a self-adaptive fitting strategy of operation state perception, and the stability, adaptability and control precision of the inverter power supply under complex working conditions are effectively improved.

Robot calibration is crucial in multi-robot cooperative systems where the inaccuracy of robots can add up and cause large errors in the final trajectory of handled parts or process tools. In this work, a two-step calibration approach is proposed based on artificial neural networks (ANNs) and definition of compensated pose for a master-slave cooperative robot system. Measuring the pose of master and slave robots at different locations in their shared workspace is required to create pairs of joint angles and output pose errors as training data. The generated data is used to train two ANN models for compensating the master-slave relative error and the master robot errors. The master-slave relative error is corrected by introducing a compensated pose for the slave robot with respect to the master robot. A neural network is then trained to predict the error parameters of the compensated pose for the joint angles of both robots as the input. The master robot is then corrected individually using another ANN model to address the absolute accuracy of the cooperative system.Measurements and simulations have been performed on a dual-robot cooperative system before and after geometric calibration. The process of cross validation is carried out to find the best network architecture for the optimal performance in correcting the robots'errors. It has been shown that even after pre-existing model-based calibration of each robot, both the absolute accuracy of the master robot and the relative tracking accuracy can be further improved by the proposed implementation of ANN calibration.

Various aspects of the present disclosure generally relate to wireless communication. Some aspects relate to artificial intelligence or machine learning (AI / ML) based channel coding, in which an AI / ML model is used to encode and / or modulate a message for transmission. Aspects described herein use an artificial neural network (ANN) which may include a number of self-attention layers. For example, the ANN may consist of self-attention layers. An ANN that includes only self-attention layers may be associated with a lower number of model parameters than other ANNs, such as ANNs that incorporate a set of transformer layers. Furthermore, the number of model parameters that define an ANN may be lower than a number of table parameters that define a codebook for a transformer-based ANN for channel coding. Thus, signaling overhead can be reduced.

Systems, methods and apparatus of vehicle control. For example, a vehicle includes: a multi-master serial bus (e.g., a controller area network (CAN) bus); electronic control units connected to the bus and configured to communicate with each other through the bus; a transceiver connected to the bus and configured to monitor communication traffic on the bus to generate inputs; and an artificial neural network configured to generate, based on the inputs, a classification of anomaly for the communication traffic on the bus. The vehicle is configured to apply a security measure in response to the classification of anomaly for the communication traffic on the bus.

The invention discloses a knowledge-driven intelligent metasurface unit rapid design method based on double neural networks. Aiming at the problems that a traditional intelligent metasurface unit design process depends on a large amount of electromagnetic simulation and is complex in calculation and low in efficiency, the invention provides a dual-artificial neural network (ANN) proxy model optimization framework combined with physical knowledge guidance. The method comprises the following steps: firstly, respectively training two neural network models to predict the amplitude and phase response of a metasurface unit of a tuning element in different extreme working states; and then, according to the physical prior knowledge of monotonic change of tuning element parameter change on electromagnetic response, the maximum reflection loss and phase difference under any geometric parameter combination are quickly and accurately deduced, so that double-target optimization design assisted by the agent model is realized. By means of the method, the design efficiency and performance of the metasurface unit can be greatly improved, dependence on traditional full-wave electromagnetic simulation is remarkably reduced, and the effectiveness and superiority of the method are verified through simulation and experimental results.

Aspects of this disclosure relate to digital compensators, such as digital predistortion systems. Digital predistortion systems disclosed herein use a neural Volterra approach. Such digital predistortion systems can include a first processing path, an envelope processing path comprising an envelope artificial neural network, multipliers configured to multiply respective output signals of the first processing path and the envelope processing path, and a combiner configured to generate a combined output signal based on at least output signals of the multipliers. The combined output signal is a digitally predistorted version of the input signal.

The invention relates to the technical field of air temperature risk assessment, in particular to an AdaBoost-based urban air temperature inference and thermal exposure risk assessment method. The method comprises the following steps: acquiring original city form data and city meteorological data of a city; performing data preprocessing on the original city form and the city meteorological data of the city to generate processed original city form data and city meteorological data; integrating the processed original city form data and city meteorological data into a model training set and a model test set; and designing an artificial neural network architecture by adopting an AdaBoost algorithm, and performing model training on the model training set by utilizing the artificial neural network architecture to generate an urban air temperature inference pre-model. According to the method, by fusing multi-source data, applying an advanced machine learning algorithm and enhancing sensitivity analysis and Monte Carlo simulation, the accuracy and reliability of urban air temperature inference and thermal exposure risk assessment are improved.