Alerta

Obstacle Detection Method and Device for Assisting Vehicle in Driving

Resumen de: US20260276813A1

An obstacle detection method for assisting in vehicle driving includes (i) obtaining ultrasonic echo data captured during vehicle movement, (ii) obtaining information associated with echo intersections based on the ultrasonic echo data, (iii) providing at least part of the ultrasonic echo data and the information associated with the echo intersections as feature data to a machine learning model to obtain detection information for an obstacle, wherein the machine learning model employs at least one of a classification algorithm or a regression algorithm, and (iv) assisting in vehicle driving based on the detection information for the obstacle.

Mitigating cyber-threats in a network

Resumen de: GB2704692A

A method for mitigating cyber-threats in a target network comprising a plurality of nodes. A decentralised multi-agent model with a plurality of local models is deployed to the target network S210. Each of the plurality of local models monitors a local region of the target network to obtain local network information S220 and predicts threat mitigation actions based on the local network information S230. The decentralised multi-agent model may be a machine learning model, trained using reinforcement learning. The local models may be trained using one or more training networks, using a trainer system which iteratively evaluates a performance of the multi-agent model and adjusts the local models. Figure 2

Machine learning with PII protection and explainability

Resumen de: GB2704709A

Processing input data 300, comprising receiving an encrypted model parameter update, wherein at least model parameters relating to personally identifiable information (PII) are encrypted using a homomorphic encryption algorithm 302. The encrypted model parameter update is decrypted using the homomorphic encryption algorithm 304. The decrypted model parameter update is applied to model parameters of a machine learning model 306. The machine learning model receives input data comprising personally identifiable information 308. A machine learning model is used to process the input data to produce output data 310. Finally, an explainability algorithm is applied to the output data 312. The receiving of the encrypted model parameter update may comprise the receiving of an encrypted model parameter update from a server device that aggregate the encrypted model parameter update from local encrypted model parameter updates received from a plurality of user devices. Fig. 3

FLEXIBLE MACHINE LEARNING MODEL COMPRESSION

Resumen de: WO2025101721A1

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for compresses a machine learning model having a plurality of parameters. In one aspect, one of the methods includes obtaining trained values of a set of parameters for at least a portion of a machine learning model; identifying one or more dense ranges for the trained values; determining a least number of bits required to represent each trained value within the one or more dense ranges; identifying a second format having a range that is smaller than a range of the first format; and generating a compressed version of the at least a portion of the machine learning model.

MACHINE LEARNING BASED SEMICONDUCTOR MEASUREMENT MODELS TRAINED USING HISTORICAL DATA

Resumen de: US20250238690A1

Methods and systems for using historical measurement data to train a present state, machine learning (ML) based measurement model are described herein. This approach takes advantage of the correlation between structural characteristics of measured samples fabricated in accordance with different design revisions, process revisions, or both. In one aspect, a present state, ML based measurement model is trained using training data associated with measurements of a plurality of instances of a current version of a semiconductor structure in a present state of a semiconductor process flow and training data associated with measurements of a plurality of instances of a historical version of the semiconductor structure in the present state of the semiconductor process flow. In some examples, training data also includes prior state measurement data. Historical training data, prior state training data, or both, may be derived from actual reference measurements, in-line, production measurements, or both.

A METHOD FOR GENERATING TEST CASES FOR TESTING A SOFTWARE PROGRAM

Resumen de: EP4807565A1

0001 The invention relates to a method (100) for generating test cases for testing a software program, comprising: - Training (101) a first machine learning model (1) for a next-token prediction based on a training data set, the training data set comprising test cases for testing the software program in a code format, - Providing (102) a labelled training data set, the labelled training data set comprising test cases that are each associated with at least one characteristic of the software program, - Performing (103) a respective fine-tuning of the trained first machine learning model (1) for the next-token prediction based on the labelled training data set for each characteristic of the software program to provide a specialized machine learning model (2) for each respective characteristic, - Generating (104) the test cases for testing the software program for at least one of the characteristics of the software program by utilizing the first machine learning model (1) and/or a respective specialized machine learning model (2) that was fine-tuned for said at least one of the characteristics.

Control parameter optimization

Resumen de: GB2704660A

A computer system comprises a parameter optimization module 102 for optimizing input parameters of a process having a binary output, the module comprising: probabilistic machine learning model 1021 trained to model a relationship between a first and a second controllable input parameter of the process and the binary output, the relationship defining an optimized pair of input parameter values 104; and batched binary Bayesian Optimizer 1022 configured to generate, from the trained probabilistic machine-learning model, a list 105 comprising training pairs of input parameter values. The parameter-optimization-module 102 is configured to output the optimized pair of input parameter values 104 and the list 105 of training pairs of input parameter values, and to receive binary output values of the process performed using the training pairs of input parameter values. The computer system is configured to periodically retrain to generate a new optimized pair 104 of input values and a new list 105 of training pairs of input parameters. Also claimed is assigning received requests to a first or second group to perform the process using the optimized or training input parameters, respectively. Utilizing batched binary Bayesian optimization creates a much smaller training data set by selecting training parameters with the greatest impact when model 1021 is retrained. The computer system operates using optimized input parameters 104 that produce desired results, yet still producing new trai

A FISH SPECIES DISTRIBUTION AND ABUNDANCE PREDICTION METHOD BASED ON ENHANCED ENSEMBLE MACHINE LEARNING

Resumen de: NL4002365A

0001 The present invention discloses a fish species distribution and abundance prediction method based on enhanced ensemble machine learning, relating to the technical field of fish ecological prediction. This method integrates the characterization capability of enhanced ensemble learning for complex environmental responses and the representation capability of hierarchical models for community ecological processes, and can effectively resolve nonlinear fixed effects, spatial dependence, and species interactions, thereby improving the accuracy and ecological interpretability of multi-species prediction.

TRAINING DEVICE, INFORMATION PROCESSING APPARATUS, SUBSTRATE PROCESSING DEVICE, TRAINING METHOD AND PROCESSING CONDITION DETERMINATION METHOD

Resumen de: US20260268050A1

A training device includes a first hardware processor, wherein the first hardware processor acquires a first dataset including a processing condition for a process to be executed by a substrate processing device, and a processing result of the process, generates a pre-processing condition, causes a learning model to execute machine learning using the second dataset, and causes the trained learning model to execute machine learning using the first dataset, with the trained learning model having executed machine learning using the second dataset, and the second dataset includes a pre-processing result that is predicted by a predetermined prediction algorithm based on the pre-processing condition, and the pre-processing condition.

METHOD AND DEVICE FOR TRANSMITTING/RECEIVING SIGNAL IN WIRELESS COMMUNICATION SYSTEM

Resumen de: US20260270934A1

A method performed by a first device in a wireless communication system, according to at least one embodiment among the embodiments disclosed in the present specification, comprises: receiving, from a second device, one or two or more data sets related to positioning; training an artificial intelligence/machine learning (AI/ML) model on the basis of at least a portion of the one or two or more data sets; and acquiring positioning information outputted from the trained AI/ML model, wherein data label-related information is given to each of the received one or two or more data sets, and the data label-related information may include positioning-related actual measurement information and information related to the quality of the actual measurement information.

LEARNING SYSTEM, PREDICTION SYSTEM, AND CONTROL SYSTEM

Resumen de: US20260268221A1

A learning system sets a mask for sensor data related to a state of a prediction target, and receives inputs of sensor data masked according to a set mask and information indicating a setting of the mask to perform training of a machine learning model that outputs a prediction value of data regarding the state of the prediction target.

SYSTEM AND METHOD FOR MULTIMODAL VERIFICATION AUTOMATED VALUE CHAIN DECARBONIZATION

Resumen de: WO2026187958A1

A computer-implemented system and method for automated value chain decarbonization management that integrates geospatial verification, machine learning anomaly detection, constrained optimization, distributed ledger recording, and cryptographic attestation into a unified processing pipeline. The integrated pipeline transforms raw, unverified supply chain data into cryptographically-verified, tamper-evident emissions documentation.

ARTIFICIAL INTELLIGENCE SENTINEL FOR SURGICAL PLANNING

Resumen de: US20260263157A1

A computer-implemented method of machine learning based surgical planning optimization. Embodiments include receiving, by an artificial intelligence agent, a request related to a surgical procedure that is to be performed on a patient. Embodiments include retrieving medical data that is related to the request from one or more source devices. Embodiments include generating, using a machine learning model, content related to the surgical procedure that is to be performed on the patient based on the request and the medical data, wherein the content comprises a set of relevant data points about the patient with respect to the surgical procedure or an indication of an issue related to a surgical plan. Embodiments include providing the content via an output device.

CONTENT GENERATION ASSISTANT

Resumen de: WO2026187951A1

Aspects of the present disclosure involves systems and techniques for content generation system. The systems and techniques include receiving disclosure materials; generating, by a claim generator comprising a machine learning model trained on claim language, one or more claims based on the disclosure materials; receiving one or more changes to the one or more claims; and inputting the one or more changes to the claim generator as training data to further train the claim generator.

Systems and Methods For Leveraging Large Language Model Explainability In Deep Learning Classification

Resumen de: US20260268184A1

0000 An explanatory engine is disclosed that explains classifications from a classification system. The explaining engine includes an investigation appliance, an enrichment agent, and an explanatory agent. The investigation appliance is configured to forward a classification result from the classification system to a user and to communicate a user communication related to the classification result. The enrichment agent is configured to generate an enrichment content for the classification result based on the user communication and classification information received from the classification system. The explanatory agent configured to transmit an explanation of the classification result based on the enriched content and the user communication.

AUTOMATIC PRUNING MASK FOR LARGE LANGUAGE MODELS WITH BALANCED WEIGHT AND ACTIVATION

Resumen de: US20260268142A1

0000 An apparatus and method for efficiently performing efficient data storage and data transfer of machine learning data. In various implementations, one of the multiple processing circuits of the computing system retrieves matrix of machine learning (ML) model weights (or weights). The processing circuit performs magnitude normalization across one or more channel dimensions of the matrix of weights. The processing circuit performs activation magnitude normalization across one of the channel dimensions of the input activation values. The processing circuit generates a pruning mask for the matrix of weights based on a combination of the weight magnitude normalization and the activation magnitude normalization.

MEDIA ITEM ANALYSIS

Resumen de: WO2026185565A1

Methods and systems are provided for generating a model output. A method comprises receiving eye-tracking data and second eye-tracking data and processing, using a machine learning model, a model input representing the eye-tracking data, the second eye-tracking data, and synchronization metrics indicating a degree of synchronization between the individual and the one or more second individuals to generate a model output. The model output is indicative of a relationship between electrical brain activity and the one or more properties of the at least one eye.

SYSTEMS AND METHODS FOR GENERATING ROLE-PLAYING AI PERSONAS CONSTRUCTED FROM VARIOUS MARKETING SEGMENTS

Resumen de: US20260268178A1

0000 Methods and systems are described for an AI persona generation system. A system can provide groups of AI personas with specific facets reflecting a desired consumer group. AI/ML tools can be used to optimize the generation of AI personas. Systems and methods include receiving, via a user interface, a request for one or more artificial-intelligence (AI) personas; determining, by the persona generation server, a plurality of key persona facets for the request; generating an AI persona mold based on the plurality of key persona facets; generating, by an artificial intelligence/machine learning (AI/ML) model, a plurality of AI personas consistent with the persona mold; and validating the plurality of AI personas based on one or more validation criteria.

METHODS AND SYSTEMS FOR SENSOR ANALYSIS OF A METABOLOME

Resumen de: WO2026187913A1

In an aspect, the present disclosure provides a method of determining a profile of a subject, comprising: (a) obtaining data obtained from a biological sample of said subject; (b) processing, using a machine-learning (ML) algorithm, said data to generate output data, wherein, prior to said processing, a dimensionality of said data is not reduced; and (c) determining said profile based at least in part on said output data.

SYSTEMS AND METHODS FOR PROBABILISTIC MODEL DEVELOPMENT WITH TRAINING AND INFERENCE ARCHITECTURES

Resumen de: WO2026187351A1

Systems and methods for probabilistic machine learning model development are disclosed, featuring distinct yet harmonized training and inference architectures. The training architecture compresses large-scale data into probability distributions and performs iterative sampling during training, reducing computational overhead while preserving statistical properties. The inference architecture leverages these learned distributions to generate multiple probabilistic predictions for robust outputs. A specialized hardware implementation includes sensor interfaces for data acquisition, accelerated computing components for training, and edge computing modules for efficient inference. The system is particularly suited for disease diagnostics, including cancer detection, where data may be high-dimensional and noisy. Unlike conventional approaches that process raw data directly, the disclosed framework provides measurable improvements in power usage, throughput, and prediction reliability through customized modules for data handling, probabilistic sampling, and model updates. The architecture enables efficient resource management across both training and inference phases while maintaining mathematical consistency.

MACHINE LEARNING MODEL ANALYSIS

Resumen de: US20260268227A1

Disclosed are various embodiments for analyzing machine learning models. A selection is obtained of a first tuple comprising a first feature vector and a first result generated by a machine learning model and a second tuple comprising a second feature vector and a second result generated by the machine learning model. Then, a plurality of emulated feature vectors are generated. Next, a plurality of emulated results are generated. Subsequently, a plurality of emulated decision instances are generated. Next, a decision tree is built based at least in part on the first tuple, the second tuple, and the plurality of emulated decision instances. Finally, an importance of each feature on the decision tree is computed.

OBTAINING THE BEST NOISE FACTORS AND EXTRAPOLATOR TO BE USED IN QUANTUM ERROR MITIGATION

Resumen de: US20260268197A1

A method, system and computer program product for performing quantum error mitigation. A machine learning model is trained to predict the optimal noise factors and optimal extrapolator to be used in quantum error mitigation based on the quantum circuits, such as the structures of the quantum circuits, and the selections of different quantum hardware (e.g., noise profile of the selected quantum hardware). Based on the received structure of a quantum circuit and the selected quantum hardware, the optimal noise factors and the optimal extrapolator to be used in quantum error mitigation for the received quantum circuit to be run on the selected quantum hardware are identified using the trained machine learning model. Quantum error mitigation is then performed on the quantum circuit, such as the received quantum circuit, after the quantum circuit has been run on the selected quantum hardware using the identified optimal noise factors and optimal extrapolator.

SYSTEM AND METHOD TO DETECT MARINE ACTIVITIES

Resumen de: US20260268418A1

0000 The present invention relates to a method for detecting a vessel engaging in illegal fishing. The method includes the steps of: receiving automatic identification system (AIS) data from transceivers of vessels; preprocessing the AIS data to generate a set of motion-related features; constructing one or more trajectory representations from the set of motion-related features; extracting one or more features from the one or more trajectory representations with an auto-encoder module; and selecting one or more random sample data for an Artificial Intelligence Engine for detecting a vessel engaging in fishing. The Artificial Intelligence Engine includes an unsupervised learning module for labeling the one or more trajectory representations into a plurality of labels; and a supervised learning module for generating a machine learning model for determining the vessel engaging in illegal fishing.

RESOLVING CELLULAR NETWORK CONNECTIVITY DEFICIENCIES USING MACHINE LEARNING-BASED TREND ANALYSIS

Resumen de: US20260268203A1

0000 A method includes executing accessibility algorithm(s) on connectivity data to determine a set of performance metrics for cellular network accessibility through network component(s). The method analyzes the performance metrics over time to detect trends and to determine correlations between network performance and access demand patterns. The method trains a ML model using, as inputs, the trends and the correlations detected within the performance metrics and monitors outputs of the ML to detect whether the outputs drop below performance threshold values for the network component(s). In response to detecting an output of the ML model drop below a performance threshold value for a network component, the method causes an increase in distributed unit or centralized unit resources associated with the network component that will increase a level of network accessibility to cellular devices through the network component.

SYSTEMS AND METHODS FOR AUTOMATED PSYCHOLOGICAL ASSESSMENT AND MENTAL HEALTH EVALUATION

Nº publicación: WO2026187714A1 10/09/2026

Solicitante:

STARK RHONDA [US]
STARK, Rhonda

Resumen de: WO2026187714A1

An Al-driven psychological assessment system automates mental health evaluations using structured assessments, machine learning diagnostics, and adaptive referrals. The system collects user inputs, analyzes responses through an Al module, and compares results against a diagnostic database incorporating DSM-V criteria and historical case data. A severity engine assigns risk scores, prioritizes findings, and generates provisional diagnoses validated against prior records. Based on assessed severity, the system recommends treatment pathways, delivers AI-guided support for non-severe conditions, and limits automated counseling for high-risk users. When elevated risk is detected, the platform initiates structured referrals and emergency interventions, connecting users with licensed mental health professionals. Through real-time data processing, continuous diagnostic refinement, and risk-sensitive safeguards, the system improves the accuracy, scalability, and accessibility of psychological assessments while maintaining patient safety via controlled escalation protocols and professional oversight.