Alerta

CHANNELWISE FEATURE REORIENTATION AND MERGING WITH NEIGHBOURING FEATURES FOR ROBUST LOW COMPLEXITY DNN BASED AUDIO PROCESSING

Absstract of: WO2025228896A1

There is disclosed an apparatus (10) for processing an input information signal (20), comprising: * a feature extractor (100) for extracting a set of features (102) from the input information signal (20), the set of features (102) having a first shape, * a feature block generator (200, CR1) configured to reshape the set of features (102) from a first shape to a second shape by: - segmenting (210) the set of features (102) into at least 3 feature subsets (212, 212a-212h), each feature subset (212, 212a-212h) having a specific shape, - performing a subsampling (220, 220a, 220b) of the feature subsets (212) to generate a plurality of feature blocks (222, 222a-222d) by stacking a selected set of the subsampled feature subsets feature subsets (212, 212a-212h) along a first dimension, - creating (230) a feature hyper-block (332) of the second shape by stacking the plurality of feature blocks (222, 222a-222d) along a second dimension, * a neural network processor (300, D1 ) configured to process the feature hyper block (332) to obtain a first result (380); and * an output post-processor (P2, 700) configured to post-process the result (380) to construct an output information signal (40) and/or to output a classification result.

DEVICE, METHOD, AND PROGRAM FOR ENHANCING OUTPUT CONTENT THROUGH ITERATIVE GENERATION

Absstract of: US2025342836A1

A method of improving output content through iterative generation is provided. The method includes receiving a natural language input, obtaining user intention information based on the natural language input by using a natural language understanding (NLU) model, setting a target area in base content based on a first user input, determining input content based on the user intention information or a second user input, generating output content related to the base content based on the input content, the target area, and the user intention information by using a neural network (NN) model, generating a caption for the output content by using an image captioning model, calculating similarity between text of the natural language input and the generated output content, and iterating generation of the output content based on the similarity.

THERMAL OPTIMIZATION AND CONTROL IN OPEN-PLAN SPACES USING PHYSICS-INFORMED GRAPH NEURAL NETWORK BASED OPTIMAL CONTROLLER

Absstract of: US2025341329A1

Optimal control of multiple heating, ventilation, and air-conditioning units in an open-plan space demands fast and accurate thermodynamic modeling. Prior methods lack scalability required for effective control in large open-plan offices primarily due to air-mixing interactions. The present disclosure describes a physics-informed graph neural network (PI-GNN) to overcome these challenges. Specifically, thermodynamic interactions are modeled as edges between nodes that represent cells. Further, a modeling approach is used that allows explicit modeling of wall and window surface temperatures which are commonly ignored. The method of present disclosure utilizes PI-GNN as a state-estimator that employs a receding-horizon approach for optimal HVAC control. PI-GNNs are adapted for building HVAC control by incorporating a time-resetting strategy to handle time-dependent ambient conditions and therefore set-points. The method of the present disclosure outperforms a regular PINN model and other baseline control strategies on thermal model accuracy, computation time, energy consumption, and user comfort.

HISTOPATHOLOGY-BASED SOLID TUMOUR ANALYSIS

Absstract of: US2025342974A1

A computer-implemented method is provided, for predicting a risk factor for a patient based on histopathological image analysis. The method includes: receiving at least one histological image of a solid tumour; converting the histological image into a graph representation; processing the graph representation using a neural network, wherein the neural network comprises a graph isomorphism network and a convolutional neural network; and determining the risk factor based on an output of the neural network. Also provided is a method of training one or more neural networks for use in such a method.

HISTOPATHOLOGY-BASED SOLID TUMOUR ANALYSIS

Absstract of: EP4645338A1

A computer-implemented method is provided, for predicting a risk factor for a patient based on histopathological image analysis. The method includes: receiving (110) at least one histological image of a solid tumour; converting (150) the histological image into a graph representation; processing (160) the graph representation using a neural network, wherein the neural network comprises a graph isomorphism network and a convolutional neural network; and determining (162) the risk factor based on an output of the neural network. Also provided is a method of training one or more neural networks for use in such a method.

Histopathology-based solid tumour analysis

Absstract of: GB2640747A

The method uses histopathology images preferably of H&E stained tissue samples collected from solid tumours of cancer patients to predict cancer relapse risk levels (known as prognosis) based on artificial intelligence technology. The method is for predicting a risk factor for a patient based on histopathological image analysis. The method includes: receiving 110 at least one histological image of a solid tumour; converting 150 the histological image into a graph representation comprising extracting 120 a region of interest from the histological image and constructing the graph representation based on the region of interest; processing 160 the graph representation using a neural network, wherein the neural network comprises a graph isomorphism network and a convolutional neural network; and determining 162 the risk factor based on an output of the neural network. Also provided is a method of training one or more neural networks for use in such a method. The training relies on a backpropagation algorithm.

CHANNELWISE FEATURE REORIENTATION AND MERGING WITH NEIGHBOURING FEATURES FOR ROBUST LOW COMPLEXITY DNN BASED AUDIO PROCESSING

Absstract of: EP4645165A1

There is disclosed an apparatus (10) for processing an input information signal (20), comprising:a feature extractor (100) for extracting a set of features (102) from the input information signal (20), the set of features (102) having a first shape,a feature block generator (200, CR1) configured to reshape the set of features (102) from a first shape to a second shape by:segmenting (210) the set of features (102) into at least 3 feature subsets (212, 212a-212h), each feature subset (212, 212a-212h) having a specific shape,performing a subsampling (220, 220a, 220b) of the feature subsets (212) to generate a plurality of feature blocks (222, 222a-222d) by stacking a selected set of the subsampled feature subsets feature subsets (212, 212a-212h) along a first dimension,creating (230) a feature hyper-block (332) of the second shape by stacking the plurality of feature blocks (222, 222a-222d) along a second dimension,a neural network processor (300, D1) configured to process the feature hyper block (332) to obtain a first result (380); andan output post-processor (P2, 700) configured to post-process the result (380) to construct an output information signal (40) and/or to output a classification result.

CODING OF AN EVENT IN AN ANALOG DATA FLOW WITH A FIRST EVENT DETECTION SPIKE AND A SECOND DELAYED SPIKE

Absstract of: US2025335754A1

A set of spike pattern data is stored on a set of storage neurons of a neural network. The storing includes storing first parameters indicative of a presence of spikes on respective neurons of the set of storage neurons, and storing, on the set of storage neurons of the neural network, second parameters indicative of a timing of spikes on the respective neurons of the set of storage neurons.

ARTIFICIAL INTELLIGENCE MODELING TO PREDICT ELECTRONIC ACCOUNT DATA

Absstract of: US2025335967A1

Disclosed methods and system describe a server that uses AI modeling to predict negative cash flow at a user level. The server periodically retrieves data associated with the user, the data comprising monetary attributes associated with one or more accounts of the user; executes a deep neural network model trained based upon historical data associated with at least a subset of the users configured to predict a negative cash flow in one or more accounts of the user, a depth of the negative cash flow, and a duration of the negative cash flow; transmits, to a second server, the predicted values, whereby when the second server determines that a likelihood of account needs satisfies a threshold, the second server establishes an electronic communication session with an electronic device of the user; trains the deep neural network when the second server establishes the electronic communication session.

SYSTEM AND METHOD FOR PREDICTING FORMATION IN SPORTS

Absstract of: US2025336208A1

A system and method of predicting a team's formation on a playing surface are disclosed herein. A computing system retrieves one or more sets of event data for a plurality of events. Each set of event data corresponds to a segment of the event. A deep neural network, such as a mixture density network, learns to predict an optimal permutation of players in each segment of the event based on the one or more sets of event data. The deep neural network learns a distribution of players for each segment based on the corresponding event data and optimal permutation of players. The computing system generates a fully trained prediction model based on the learning. The computing system receives target event data corresponding to a target event. The computing system generates, via the trained prediction model, an expected position of each player based on the target event data.

SYSTEMS AND METHODS FOR COMPUTING FEATURING SYNTHETIC COMPUTING OPERATORS AND COLLABORATION

Absstract of: WO2025226986A1

One embodiment is directed to a synthetic engagement system for process-based problem solving, comprising: a computing system comprising one or more operatively coupled computing resources; and a user interface operated by the computing system and configured to engage a human operator in accordance with a predetermined process configuration toward an established requirement based at least in part upon one or more specific facts; wherein the user interface is configured to allow the human operator to select and interactively engage one or more synthetic operators operated by the computing system to proceed through the predetermined process configuration, and to return result to the human operator selected to at least partially satisfy the established requirement; and wherein each of the one or more synthetic operators is informed by a convolutional neural network informed at least in part by historical actions of a particular actual human operator and a synthetic operator background configuration.

Method and Apparatus for Inferencing By Fusing Pre-processing into Neural Network

Absstract of: KR20250152434A

본 실시예는 신경망에 전처리 기능을 융합한 추론방법 및 장치를 개시한다. 본 실시예에서, 추론 장치는, 다수의 레이어들을 포함하는 딥러닝 기반 신경망을 이용하여 입력 영상으로부터 입력 영상의 특징을 생성하고, 입력 영상의 특징에 기초하여 추론 결과를 생성한다. 신경망의 첫 번째 레이어는 신경망의 트레이닝 과정에서 적용된 입력 정규화를 융합하기 위해, 입력 정규화에 사용된 인자들에 기초하여 정규화된 가중치들 및 바이어스를 포함한다. 신경망의 첫 번째 레이어는 정규화된 가중치들 및 바이어스를 이용하여 입력 영상을 처리한다. 이때, 입력 정규화를 인자들은, 스케일링 인자, 평균 및 표준편차를 포함하고, 트레이닝 과정에서 활용되는 영상 데이터 전체로부터 산정된다.

Semantic Segmentation to Identify and Treat Plants in a Field and Verify the Plant Treatments

Absstract of: US2025328760A1

A farming machine including a number of treatment mechanisms treats plants according to a treatment plan as the farming machine moves through the field. The control system of the farming machine executes a plant identification model configured to identify plants in the field for treatment. The control system generates a treatment map identifying which treatment mechanisms to actuate to treat the plants in the field. To generate a treatment map, the farming machine captures an image of plants, processes the image to identify plants, and generates a treatment map. The plant identification model can be a convolutional neural network having an input layer, an identification layer, and an output layer. The input layer has the dimensionality of the image, the identification layer has a greatly reduced dimensionality, and the output layer has the dimensionality of the treatment mechanisms.

DYNAMIC DETERMINATION OF INFERENCE-TIME PARAMETERS

Absstract of: EP4636650A1

Some embodiments are directed to a method for dynamic determination of inference-time parameters to control the stochastic generation process of a generative neural network. The method may include dynamically determining for an inference request, at least from operational context information, at least one of the inference-time parameters.

Method And Apparatus for Optimizing Process Variables Based on Neural Network Inversion Algorithm

Absstract of: KR20250151661A

신경망 반전 알고리즘 기반 공정 변수 최적화를 위한 방법 및 장치를 개시한다. 본 개시의 일 측면에 의하면, 공정 변수의 최적화를 위한 컴퓨터 구현 방법으로서, 복수의 공정 변수 각각에 대한 초기 값을 포함하는 공정 변수 데이터를 획득하는 과정; 물성 예측 모델을 이용하여 상기 공정 변수 데이터로부터 물성을 예측하는 과정; 상기 예측된 물성과 목표 물성을 비교하여 손실 값을 계산하는 과정; 상기 복수의 공정 변수 각각에 대한 상기 손실 값의 편미분 값을 상기 복수의 공정 변수 각각의 그래디언트로 산출하는 과정; 공정 변수 별로 산출된 그래디언트 및 기 설정된 제약에 기초하여 상기 공정 변수 데이터를 갱신하는 과정; 및 상기 손실 값이 기 정의된 임계값 미만이 될 때까지 상기 물성을 예측하는 과정부터 상기 공정 변수 데이터를 갱신하는 과정까지를 반복 수행하는 과정을 포함하는 방법을 제공한다.

NEURAL NETWORK CLASSIFIERS FOR BLOCK CHAIN DATA STRUCTURES

Absstract of: US2025323782A1

Disclosed is a neural network enabled interface server and blockchain interface establishing a blockchain network implementing event detection, tracking and management for rule based compliance, with significant implications for anomaly detection, resolution and safety and compliance reporting.

DYNAMIC DETERMINATION OF INFERENCE-TIME PARAMETERS

Absstract of: US2025322212A1

A method for dynamic determination of inference-time parameters to control the stochastic generation process of a generative neural network. The method may include dynamically determining for an inference request, at least from operational context information, at least one of the inference-time parameters.

CODE UNIT GENERATOR FOR A MACHINE LEARNING BASED QUESTION AND ANSWER (Q&A) ASSISTANT

Absstract of: US2025322271A1

A multimodal content management system having a block-based data structure can include an artificial intelligence (AI)-based code unit generator that can generate code units executable against the block-based data structure to provide information requested by users. For example, the code units can be generated in response to natural language prompts received via a question and answer Q&A assistant engine. A neural network can be trained on block types, block dependencies, block content values, block content types, and/or block format. The neural network can receive a set of tokens generated based on a natural language prompt and generate one or more query strings to be included in a particular code unit. The tokens can be indicative of block properties, content, or other items in the block-based data structure. The code unit can be structured to execute more than one query against the block-based data structure such that a particular result set can include content items of different modalities.

TOKEN SELECTION IN TRANSFORMER NEURAL NETWORKS FOR EFFICIENT INFERENCING

Absstract of: US2025322275A1

Certain aspects of the present disclosure provide techniques and apparatus for processing data using a transformer neural network. The method generally includes generating, via a first attention layer of a machine learning model, a first attention map based on an input into the machine learning model; identifying, using a token prediction model, a first subset of tokens in the first attention map more relevant to a second attention layer of the machine learning model and a second subset of tokens in the first attention map less relevant to the second attention layer of the machine learning model; generating, via the second attention layer of the machine learning model, a second attention map based on the first subset of tokens in the first attention map; and generating an inference based on the second attention map and the second subset of tokens in the first attention map.

METHOD, SYSTEM, AND COMPUTER PROGRAM PRODUCT FOR GLOBAL PERSONALIZED RECOMMENDATION

Absstract of: WO2025217351A1

Methods, systems, and computer program products for providing global personalized recommendations are provided. An example method may include generating embeddings for a first plurality of entities based on a first dataset, determining first identifiers of the first plurality of entities included in the first dataset that corresponds to second identifiers of a second plurality of entities included in a second dataset to provide a matched set of entities, wherein the second dataset includes attribute data associated with each entity of the second plurality of entities, generating a graph representation of the second plurality of entities, and wherein the graph includes nodes and each node represents an entity of the second plurality of entities, determining one or more first nodes that lacks data associated with a node embedding, and generating data associated with the node embedding for the one or more first nodes using a graph neural network (GNN) machine learning model.

SYSTEM AND METHOD FOR VARIATIONAL ANNEALING TO SOLVE FINANCIAL OPTIMIZATION PROBLEMS

Absstract of: AU2024204682A1

A system and method for variational annealing to solve financial optimization problems is provided. The financial optimization problem is encoded as objective function represented in terms of an energy function. An autoregressive neural network is trained to minimize the cost function via variational emulation of classical or quantum annealing. Optimal solutions to the financial optimization problem are obtained after a stopping criterion is set. An optimal solution may be selected according to user defined metrics, and optionally applied to a real-world system associated with the financial optimization problem. A system and method for variational annealing to solve financial optimization problems is provided. The financial optimization problem is encoded as objective function represented in terms of an energy function. An autoregressive neural network is trained to minimize the cost function via variational emulation of classical or quantum annealing. Optimal solutions to the financial optimization problem are obtained after a stopping criterion is set. An optimal solution may be selected according to user defined metrics, and optionally applied to a real-world system associated with the financial optimization problem. ul u l s y s t e m a n d m e t h o d f o r v a r i a t i o n a l a n n e a l i n g t o s o l v e f i n a n c i a l o p t i m i z a t i o n p r o b l e m s i s p r o v i d e d h e f i n a n c i a l o p t i m i z a t i o n p r o b l e m i s e n c o d e d a s o b

AUDIO NOISE DETERMINATION USING ONE OR MORE NEURAL NETWORKS

Absstract of: US2025324199A1

Apparatuses, systems, and techniques are presented to reduce noise in audio. In at least one embodiment, one or more neural networks are used to determine a noise signal in one or more speech signals.

REAL TIME CONTEXT DEPENDENT DEEP LEARNING

Absstract of: US2025322233A1

In an example, an apparatus comprises a plurality of execution units comprising and logic, at least partially including hardware logic, to receive a plurality of data inputs for training a neural network, wherein the data inputs comprise training data and weights inputs; represent the data inputs in a first form; and represent the weight inputs in a second form. Other embodiments are also disclosed and claimed.

GRAPH-BASED MODELING OF RELATIONAL AFFECT IN GROUP INTERACTIONS

Absstract of: US2025322206A1

According to one aspect, graph-based modeling of relational affect in group interactions may include generating a graph neural network (GNN) based on multi-modal behavioral data associated with interactions between two or more individuals for each of the two or more individuals and relational context information associated with the interaction or the two or more individuals, performing message passing between nodes of the GNN based on the relational context information, generating a representation read-out associated with the GNN or a subgraph of the GNN, and performing an action based on the representation read-out.

AUGMENTING MACHINE LEARNING LANGUAGE MODELS USING SEARCH ENGINE RESULTS

Nº publicación: US2025322236A1 16/10/2025

Applicant:

GDM HOLDING LLC [US]
GDM Holding LLC

Absstract of: US2025322236A1

Methods, systems, and apparatus, including computer programs encoded on computer storage media, for augmenting machine learning language models using search engine results. One of the methods includes obtaining question data representing a question; generating, from the question data, a search engine query for a search engine; obtaining a plurality of documents identified by the search engine in response to processing the search engine query; generating, from the plurality of documents, a plurality of conditioning inputs each representing at least a portion of one or more of the obtained documents; for each of a plurality of the generated conditioning inputs, processing a network input generated from (i) the question data and (ii) the conditioning input using a neural network to generate a network output representing a candidate answer to the question; and generating, from the network outputs representing respective candidate answers, answer data representing a final answer to the question.