Xarxes Neuronals

Always-On Keyword Detector

Resumen de: US2025131920A1

Provided herein is an integrated circuit including, in some embodiments, a special-purpose host processor, a neuromorphic co-processor, and a communications interface between the host processor and the co-processor configured to transmit information therebetween. The special-purpose host processor is operable as a stand-alone host processor. The neuromorphic co-processor includes an artificial neural network. The co-processor is configured to enhance special-purpose processing of the host processor through the artificial neural network. In such embodiments, the host processor is a keyword identifier processor configured to transmit one or more detected words to the co-processor over the communications interface. The co-processor is configured to transmit recognized words, or other sounds, to the host processor.

SELECTION-INFERENCE NEURAL NETWORK SYSTEMS

Resumen de: WO2024047108A1

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating a response to a query input using a selection-inference neural network.

OPTIMIZING ALGORITHMS FOR TARGET PROCESSORS USING REPRESENTATION NEURAL NETWORKS

Resumen de: WO2024018065A1

Methods, systems, and apparatus, including computer programs encoded on computer storage media, for optimizing a target algorithm using a state representation neural network.

SYSTEM ON A CHIP WITH DEEP LEARNING ACCELERATOR AND RANDOM ACCESS MEMORY

Resumen de: US2025117659A1

Systems, devices, and methods related to a deep learning accelerator and memory are described. An integrated circuit may be configured with: a central processing unit, a deep learning accelerator configured to execute instructions with matrix operands; random access memory configured to store first instructions of an artificial neural network executable by the deep learning accelerator and second instructions of an application executable by the central processing unit; one or connections among the random access memory, the deep learning accelerator and the central processing unit; and an input/output interface to an external peripheral bus. While the deep learning accelerator is executing the first instructions to convert sensor data according to the artificial neural network to inference results, the central processing unit may execute the application that uses inference results from the artificial neural network.

SELF-SUPERVISED LEARNING FOR AUDIO PROCESSING

Resumen de: US2025118291A1

Methods, computer systems, and apparatus, including computer programs encoded on computer storage media, for training an audio-processing neural network that includes at least (1) a first encoder network having a first set of encoder network parameters and (2) a decoder network having a set of decoder network parameters. The system obtains a set of un-labeled audio data segments, and generates, from the set of un-labeled audio data segments, a set of encoder training examples. The system performs training of a second encoder neural network that includes at least the first encoder neural network on the set of generated encoder training examples. The system also obtains one or more labeled training examples, and performs training of the audio-processing neural network on the labeled training examples.

HIGH AVAILABILITY AI VIA A PROGRAMMABLE NETWORK INTERFACE DEVICE

Resumen de: US2025117673A1

Techniques described herein address the above challenges that arise when using host executed software to manage vector databases by providing a vector database accelerator and shard management offload logic that is implemented within hardware and by software executed on device processors and programmable data planes of a programmable network interface device. In one embodiment, a programmable network interface device includes infrastructure management circuitry configured to facilitate data access for a neural network inference engine having a distributed data model via dynamic management of a node associated with the neural network inference engine, the node including a database shard of a vector database.

METHOD, MEDIUM, AND SYSTEM FOR INTELLIGENT ONLINE PERSONAL ASSISTANT WITH IMAGE TEXT LOCALIZATION

Resumen de: US2025117839A1

Systems, methods, and computer program products for identifying a candidate product in an electronic marketplace based on a visual comparison between candidate product image visual text content and input query image visual text content. Unlike conventional optical character recognition (OCR) based systems, embodiments automatically localize and isolate portions of a candidate product image and an input query image that each contain visual text content, and calculate a visual similarity measure between the respective portions. A trained neural network may be re-trained to more effectively find visual text content by using the localized and isolated visual text content portions as additional ground truths. The visual similarity measure serves as a visual search result score for the candidate product. Any number of images of any number of candidate products may be compared to an input query image to enable text-in-image based product searching without resorting to conventional OCR techniques.

COMPUTE OPTIMIZATIONS FOR LOW PRECISION MACHINE LEARNING OPERATIONS

Resumen de: US2025117874A1

One embodiment provides an apparatus comprising a memory stack including multiple memory dies and a parallel processor including a plurality of multiprocessors. Each multiprocessor has a single instruction, multiple thread (SIMT) architecture, the parallel processor coupled to the memory stack via one or more memory interfaces. At least one multiprocessor comprises a multiply-accumulate circuit to perform multiply-accumulate operations on matrix data in a stage of a neural network implementation to produce a result matrix comprising a plurality of matrix data elements at a first precision, precision tracking logic to evaluate metrics associated with the matrix data elements and indicate if an optimization is to be performed for representing data at a second stage of the neural network implementation, and a numerical transform unit to dynamically perform a numerical transform operation on the matrix data elements based on the indication to produce transformed matrix data elements at a second precision.

CONTROL SYSTEM WITH OPTIMIZATION OF NEURAL NETWORK PREDICTOR

Resumen de: AU2023280790A1

A predictive control system includes controllable equipment and a controller. The controller is configured to use a neural network model to predict values of controlled variables predicted to result from operating the controllable equipment in accordance with corresponding values of manipulated variables, use the values of the controlled variables predicted by the neural network model to evaluate an objective function that defines a control objective as a function of at least the controlled variables, perform a predictive optimization process to generate optimal values of the manipulated variables for a plurality of time steps in an optimization period using the neural network model and the objective function, and operate the controllable equipment by providing the controllable equipment with control signals based on the optimal values of the manipulated variables generated by performing the predictive optimization process.

Knowledge-Driven Recommendation System for Conversation - Ontology and Taxonomy Binding for Recommendation System

Resumen de: US2025111193A1

A knowledge-driven recommendation system comprises a processor and a memory with computer code instructions. The executed code instructions cause the system to receive a user query, extract a topic from the query, and submit the topic and query to a neural network. The instructions may further cause the system to return, from the neural network, a collection of taxonomy and ontology pairs, and use the pairs to select information that expands on the query and topic. The taxonomy and ontology pairs are the closest matched pairs from a knowledge graph. The closest matched pairs are retrieved when the input taxonomy topic semantically matches closest to a taxonomy topic from the custom neural network, the input ontology semantically matches closest to an ontology from the custom neural network, and the taxonomy topic from the custom neural network matches closest to one of the entities in the ontology of the neural network.

PLACEMENT OF COMPUTE AND MEMORY FOR ACCELERATED DEEP LEARNING

Resumen de: US2025110808A1

Techniques in placement of compute and memory for accelerated deep learning provide improvements in one or more of accuracy, performance, and energy efficiency. An array of processing elements comprising a portion of a neural network accelerator performs flow-based computations on wavelets of data. Each processing element comprises a compute element to execute programmed instructions using the data and a router to route the wavelets. The routing is in accordance with virtual channel specifiers of the wavelets and controlled by routing configuration information of the router. A software stack determines placement of compute resources and memory resources based on a description of a neural network. The determined placement is used to configure the routers including usage of the respective colors. The determined placement is used to configure the compute elements including the respective programmed instructions each is configured to execute.

ELECTRONIC DEVICE COMPRISING DEEP NEURAL NETWORK MODEL AND OPERATION METHOD THEREFOR

Resumen de: WO2025071378A1

Provided are: an electronic device configured to obtain output data having a high bit resolution for 8-bit quantized input data through a deep neural network operation using a digital signal processor (or an artificial intelligence dedicated processor) for performing an 8-bit operation; and an operation method therefor. The electronic device can: quantize a pixel value of an image obtained through a camera to eight bits; input the quantized 8-bit image to a deep neural network model of a digital signal processor for performing an 8-bit operation; obtain first output data having an 8-bit output value, by performing inference by the deep neural network model; obtain 8-bit second output data having a bin number, within a preset output value range; and obtain an output image by combining the first output data and the second output data.

PARTICLE-BASED SIMULATORS OF PHYSICAL ENVIRONMENTS

Resumen de: WO2025068599A1

A system that uses a graph neural network to determine a representation of a physical environment at a new time step The new time step can be before or after a current time step based on representations of the physical environment at the current time step and one or more other time steps, e.g. one or more time steps before and/or after the current time step. The representation of the physical environment at the new time step may, for example, be used to generate an image of the physical environment at the new time step. The system can be used for controlling a robot interacting with the physical environment. Some examples of the techniques are specifically adapted for implementation using hardware accelerator units.

APPARATUS, METHOD AND STORAGE MEDIUM FOR IMAGE ENCODING/DECODING TO PERFORM INTRA PREDICTION USING ARTIFICIAL NEUTRAL NETWORK

Resumen de: US2025113055A1

Disclosed herein are a method, an apparatus, and a storage medium for image encoding/decoding. An intra-prediction mode for the target block is derived, and intra-prediction for the target block that uses the derived intra-prediction mode is performed. The intra-prediction mode for the target block is derived using an artificial neural network, and an MPM list for the target block is derived using information about the target block, pieces of information about blocks adjacent to the target block, and the artificial neural network. The artificial neural network outputs one or more available intra-prediction modes. Further, the artificial neural network outputs match probabilities for one or more candidate intra-prediction modes, and each of the match probabilities for the candidate intra-prediction modes indicates a probability that the corresponding candidate intra-prediction mode matches the intra-prediction mode for the target block.

METHOD, INFORMATION PROCESSING DEVICE, AND RECORDING MEDIUM STORING INSTRUCTIONS FOR SUPPORTING EVALUATION OF LEARNING PROCESS OF PREDICTION MODEL

Resumen de: US2025103881A1

A method is executed by an information processing device for supporting evaluation of a learning process of a prediction model. The method includes: training a neural network model that includes an input layer, an intermediate layer, and an output layer, based on actual data including an explanatory factor and an objective factor; and outputting statistical information on input values that are input to the intermediate layer and the output layer of the neural network model.

NON-LINEAR ENCODING AND DECODING FOR RELIABLE WIRELESS COMMUNICATION

Resumen de: US2025105859A1

In some embodiments, a method of encoding a set of information bits to produce a codeword that encodes the set of information bits for reliable communication is provided. The set of information bits is received. The set of information bits are provided to a plurality of permutation layers separated by neural network processing layers. Each permutation layer accepts an input vector and generates a reordered output vector that is a reordering of the input vector. Each neural network processing layer accepts a vector of input values and generates a vector of output values based on a non-linear function of the vector of input values. The reordered output vector of a final permutation layer of the plurality of permutation layers is provided as the codeword. In some embodiments, a corresponding method of decoding a codeword to retrieve a set of information bits is provided.

FIDELITY-BASED EXPLANABILITY FOR GNNS

Resumen de: US2025103866A1

Methods and systems include processing an input graph using a graph neural network (GNN) to generate an output. An explanation sub-graph is generated using an explainer that identifies parts of the input graph that most influence the output. A fidelity measure of the explanation sub-graph is determined that is robust against distribution shifts. An action is performed responsive to the output, the explanation sub-graph, and the fidelity measure.

HIGH-DIMENSIONAL COMPUTING BASED TRAINING AND INFERENCING

Resumen de: US2025103849A1

An embodiment establishes a neural network that comprises a plurality of layers. The embodiment receives a plurality of input data sequences into a layer of the neural network, the plurality of input data sequences comprises a first input data sequence and a second input data sequence. The embodiment superposes the first input data sequence and the second input data sequence, thereby creating a superposed embedding. The embodiment transforms the superposed embedding by applying a function to the superposed embedding, thereby creating a transformed superposed embedding. The embodiment infers a first output data element corresponding to the first input data sequence and a second output data element corresponding to the second input data sequence via application of an unbinding operation on the transformed superposed embedding.

IMPACT FUNCTION GENERATOR FOR GEOSPATIAL CLIMATE HAZARDS

Resumen de: US2025103852A1

An embodiment for generating impact functions for geospatial climate hazards based on user interactions. The embodiment may receive input data associated with a target geospatial climate hazard and a corresponding asset, the input data including one or more of a first dataset corresponding to a predetermined set of prompts, and a second dataset corresponding to a natural language exchange. The embodiment may generate, based on the first dataset an entity knowledge graph including a series of candidate variables. The embodiment may generate, based on the second dataset, a universal knowledge graph including a series of candidate function formulas. The embodiment may generate, using a graph neural network, embeddings corresponding to the entity knowledge graph and the universal knowledge graph respectively. The embodiment may perform symbolic regression, using the embeddings, to generate one or more impact functions for the target geospatial climate hazard and the corresponding asset.

PERFORMANCE OPTIMIZATION METHOD AND APPARATUS FOR TRAINING MIXTURE-OF-EXPERTS MODEL

Resumen de: US2025103922A1

The present disclosure provides a performance optimization method and apparatus for training mixture-of-experts model, which relate to the technical field of neural networks. The method includes: judging, before one iterative calculation and for each of all experts in a mixture-of-experts model, whether a current expert needs to be set as a shadow expert, and if yes, adding the current expert to a shadow expert set, and continuing to judging whether a next expert is set as a shadow expert until all the experts are judged. The present disclosure is capable of improving the speed and efficiency of training the mixture-of-experts model, and reduce the resources consumed in the mixture-of-experts model during training.

ENERGY-EFFICIENT ANOMALY DETECTION AND INFERENCE ON EMBEDDED SYSTEMS

Resumen de: US2025103898A1

A method of anomaly detection and energy-efficient inference determination includes receiving an input. A set of features of the input are extracted using an artificial neural network (ANN) to generate a latent representation of the input. A reconstruction of the input is generated using the ANN, based on the latent representation. A reconstruction error is computed based on the generated reconstruction and the input. The reconstruction error is compared to a predefined threshold to determine whether the in-distribution data or out-of-distribution data. An anomaly is detected in response to an out-of-distribution determination. A decision model is provided with the latent representation in response to the input being determined to be in-distribution data. In turn, the decision model computes an inference based on the latent representation.

METHOD AND SYSTEM FOR NEUROPSYCHOLOGICAL PERFORMANCE TEST

Resumen de: US2025104873A1

The present invention relates to a method and system for neuropsychological performance test, comprising A terminal device (101), used to interact with a cloud server (102) which stores user information and is logged into by user through the terminal device (101); the user information obtained by the terminal device (101) are input and stored to the cloud server (102) in a login state; a test module (400) comprises the user information, which is stored in the cloud server (102) or can be downloaded from the cloud server (102), and is directly accessed through said terminal device (101) and is trained by artificial neural network; and said user information comprises user biometrics or emotions identification information, neuropsychological performance test answers information, and user latency or user chronometrics information; and said terminal device (101) displays neuropsychological performance test results.

AUDIO TYPE DETECTION

Resumen de: US2025104698A1

Artificial intelligence-based processing can be used to classify audio information received from an audio input unit. In an example, audio information can be received from a microphone configured to monitor an environment. A processor circuit can identify identifying one or more features of the audio information received from the microphone and use a first applied machine learning algorithm to analyze the one or more features and determine whether the audio information includes an indication of an abnormal event in the environment. In an example, the processor circuit can use a different second applied machine learning algorithm, such as a neural network-based deep learning algorithm, to analyze the same one or more features and classify the audio information as including an indication of a particular event type in the environment.

BUS FOR TRANSPORTING OUTPUT VALUES OF NEURAL NETWORK LAYER

Resumen de: US2025103341A1

Some embodiments provide a neural network inference circuit (NNIC) for executing a neural network that includes multiple computation nodes at multiple layers. The NNIC includes multiple core circuits including memories for storing input values for the computation nodes. The NNIC includes a set of post-processing circuits for computing output values of the computation nodes. The output values for a first layer are for storage in the core circuits as input values for a second layer. The NNIC includes an output bus that connects the post-processing circuits to the core circuits. The output bus is for (i) receiving a set of output values from the post-processing circuits, (ii) transporting the output values of the set to the core circuits based on configuration data specifying a core circuit at which each of the output values is to be stored, and (iii) aligning the output values for storage in the core circuits.

ASSESSMENT OF MACHINE-LEARNING ALGORITHM

Nº publicación: EP4528555A1 26/03/2025

Solicitante:

SIEMENS MOBILITY GMBH [DE]
Siemens Mobility GmbH

Resumen de: EP4528555A1

Techniques are disclosed that enable assessment of the processing of datasets in a machine-learning algorithm such as a deep convolutional neural network. Layer-based assessment is possible. The processing of images can be assessed. Explainable artificial intelligence is possible. Safe control of autonomous vehicles is possible.