Alerta

Self-Explaining Decision Architecture and Methods of Use

Resumen de: US20260252951A1

A Self-Explaining Decision Architecture (SEDA) for machine learning-based decision-making systems capable of generating intuitive explanations for its decisions in real time. SEDA makes use of a feature extraction subsystem and a sequence interpretation subsystem to identify patterns in data followed by a decision generation subsystem that determines appropriate actions based on those patterns. Internal state information from each of these subsystems is used to generate explanations of the system's decisions. Using this information to create explanations provides insight as to the data elements the system focused on when making decisions as well as the reasoning that was used. In at least one embodiment the system uses deep learning components including a combined convolutional neural network and long short-term memory network with attention mechanisms.

METHOD FOR MOTION PLANNING FOR A VEHICLE AND METHOD FOR TRAINING A MACHINE LEARNING ALGORITHM FOR INFERRING MANEUVER COSTS IN THE CONTEXT OF MOTION PLANNING FOR A VEHICLE

Resumen de: WO2026175619A1

The invention relates to a computer-implemented method (2) for motion planning for a vehicle (1), wherein the following steps are carried out in each of a plurality of calculation cycles: providing (21) state information relating to the vehicle (1) and environment information relating to an environment of the vehicle (1); defining (22) multiple possible maneuvers (M1, M2, M3, M4) in view of the state information and the environment information; generating (23) a particular feature vector for each maneuver (M1, M2, M3, M4); on the basis of the particular feature vector, inferring (24) costs for each maneuver (M1, M2, M3, M4) with respect to an optimization problem which comprises one or more trajectory-based cost functionals, by means of a machine learning algorithm trained for this purpose; and, according to the inferred costs, selecting (25) one or more of the maneuvers (M1, M2, M3, M4), preferably together with a specific sequence, for a detailed cost evaluation on the basis of the optimization problem.

SYSTEMS AND METHODS FOR ONTOLOGY-ORIENTED GRAPH DATASET MODELING, DATA RETRIEVAL, AND HYBRID STRUCTURING OF GENERATIVE ARTIFICIAL INTELLIGENCE OUTPUTS

Resumen de: WO2026175686A1

The present disclosure relates to systems and methods for ontology-oriented graph dataset modeling, data retrieval, and hybrid structuring of generative machine learning outputs, which address the challenges of querying and extracting clinical insights from large, complex datasets. By combining knowledge graphs and large language models, embodiments of the present disclosure enable lay users to perform natural language queries on large, structured datasets and receive evidence-backed results with visual context. Embodiments include an orchestrator module powered by one or more LLMs and an instruction prompt, which selects an appropriate context retrieval tool based on a user's query. The context retrieval tool retrieves results from an ontologically-oriented graph, generates a knowledge graph, and provides a response to the query. This approach improves the efficiency and accuracy, allowing users to discover novel and unknown relationships within the dataset without needing comprehensive schema knowledge, thereby overcoming the high barrier of entry for extracting insights.

BIAS REDUCTION FOR MACHINE LEARNING FEATURE SELECTION

Resumen de: WO2026178346A1

A method may include splitting the dataset into holdout sets. Multiple training and testing sets may be generated by splitting, for each holdout fold in the holdout sets, a remaining plurality of samples not included in the holdout fold into a training and testing set. Multiple iterations of computational feature selection may be performed using the holdout sets and the training and testing sets. Each iteration may include training and validating a first instance of a machine learning model on a training and testing set before training, using the same samples in the training and testing set, a second instance of the machine learning model to operate on a subset of features selected during the training. Each iteration may further include validating of the first instance of the machine learning model before applying the second instance of the machine learning model to a holdout fold from the holdout sets.

CONTINUOUS-FLOW DYNAMIC TITRATION CALORIMETER INCORPORATING MICROFLUIDICS AND MACHINE LEARNING-BASED CORRECTION, AND METHOD FOR USING SAME

Resumen de: WO2026174682A1

The present invention provides a continuous-flow dynamic titration calorimeter incorporating microfluidics and machine learning-based correction, and a method for using same. An apparatus comprises: a thermal insulation apparatus, used for accommodating a calorimetric structure, performing thermal insulation on the calorimetric structure, collecting the ambient temperature of the calorimetric structure, and transmitting the ambient temperature to a control and signal processing system; and the calorimetric structure, comprising a material preheating aluminum block and a constant-temperature aluminum block that are arranged at the left and right sides, wherein a material inlet/outlet connection block A is provided in a front slot of the material preheating aluminum block, a material inlet/outlet connection block B is provided in a rear slot of the material preheating aluminum block, a thermoelectric generator sheet A is provided at a front part of the constant-temperature aluminum block, a thermoelectric generator sheet B is provided at a rear part of the constant-temperature aluminum block, and a thin film heating sheet is provided above the thermoelectric generator sheet A. The technical solution of the present invention integrates microfluidics, thermoelectric power-generation measurement, and machine learning-based correction, and provides a low-cost, miniaturized and high-precision continuous-flow dynamic titration calorimetry method.

MACHINE LEARNING WITH PII PROTECTION AND EXPLAINBILITY

Resumen de: EP4797139A1

A computer-implemented method of processing input data, comprising receiving an encrypted model parameter update, wherein at least model parameters relating to personally identifiable information are encrypted using a homomorphic encryption algorithm, decrypting the encrypted model parameter update using the homomorphic encryption algorithm, applying the decrypted model parameter update to model parameters of a machine learning model, receiving, by the machine learning model, input data comprising personally identifiable information; processing, using a machine learning model, the input data to produce output data; and applying an explainability algorithm to the output data.

CONTROL PARAMETER OPTIMIZATION

Resumen de: EP4797015A1

0001 The disclosure relates to computer systems 10 and computer-implemented methods 200, 300 for performing a process having a binary output value and for optimizing controllable input parameters of such processes. One computer system 10 comprises at least one processor 11 and memory 12 configured to implement a parameter optimization module 102 for optimizing input parameters of a process having a binary output value, comprising: a probabilistic machine learning model 1021 trained to model a relationship between at least a first controllable input parameter of the process, a second controllable input parameter of the process and the binary output value of the process, the relationship defining an optimized pair of input parameter values 104 comprising an optimized value for the first input parameter and an optimized value for the second input parameter; and a batched binary Bayesian Optimizer 1022 configured to generate, from the trained probabilistic machine learning model 1021, a list 105 comprising a plurality of training pairs of input parameter values, each training pair comprising a value for the first input parameter and a value for the second input parameter. 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 10 i

AIDING MACHINE LEARNING INFORMATION RETRIEVAL BY SYMBOLIC KNOWLEDGE REPRESENTATION

Resumen de: EP4797163A1

A computer-implemented method (100) for retrieving a response (3) to a query (1) from a machine learning/artificial intelligence model, ML/Al model (2), the method (100) comprising the steps of:• providing (110) the query (1) to the ML/Al model (2), thereby obtaining an initial response (3);• determining (120), from the initial response (3), instances of concepts (5*) of a given symbolic knowledge representation (4) that the initial response (3) relates to;• marking (130), in the symbolic knowledge representation (4), each concept (5*) that the initial response (3) relates to as instantiated;• determining (140), by a reasoning engine (6), concepts (5#) of the symbolic knowledge representation (4) that, given the set of presently instantiated concepts (5*), need to be instantiated as well;• determining (150) a supplemental query (1*) for information relating at least one concept (5#) that needs to be instantiated as well; and• providing (160) this supplemental query (1*) to the ML/Al model (2), thereby obtaining a supplemental response (3*) that augments the initial response (3).

METHOD AND RELATED DEVICE FOR ACQUIRING MATHEMATICAL MODEL, AND OPERATIONS OPTIMIZATION METHOD

Resumen de: EP4797146A1

0001 A mathematical model obtaining method and device, and an operations optimization method are provided. In the method, an artificial intelligence technology may be used to obtain a mathematical model corresponding to an operations optimization problem. The method includes: outputting at least one problem, and obtaining an answer to each problem, where the answer to each problem is for obtaining first description information, the first description information is description information for describing the operations optimization problem, and a first problem is a problem for obtaining the description information of the operations optimization problem; and obtaining first information based on the first description information, and inputting the first information into a machine learning model, to obtain a mathematical model, where the mathematical model is for solving the operations optimization problem, and the mathematical model includes an objective function and a constraint. The solution greatly reduces manpower costs consumed in a process of obtaining the mathematical model, and can be adapted to obtaining mathematical models in various domains, and has high generalization.

INFORMATION PROCESSING METHOD AND RELATED DEVICE

Resumen de: EP4796407A1

Embodiments of this application disclose an information processing method and a related device. The method may be applied to the autonomous driving field in artificial intelligence. The method includes: inputting, to a deep learning model, first information corresponding to a traffic scene around an ego vehicle, and obtaining second information corresponding to the first information, where the first information includes first text information, the second information is obtained based on the deep learning model, and the second information corresponds to any one of the following tasks: making a decision on behavior of the ego vehicle, planning a trajectory for the ego vehicle, or controlling the ego vehicle. Input information of the deep learning model provided in this application includes text information that facilitates understanding by a user. Therefore, interpretability of a running process of the deep learning model is improved, to be specific, a decision-making, trajectory planning, or control process of an autonomous vehicle is more transparent, so that the user can more intuitively understand behavior of the autonomous vehicle.

MODEL MANAGEMENT METHOD AND APPARATUS

Resumen de: EP4797760A1

0001 This application provides a model management method and an apparatus, and relates to the field of communication technologies, to reduce a data security risk of an AI model in the wireless field. The method includes: obtaining a first communication data set used for updating a first machine learning model, and sending a second communication data set to a model training function entity. The first processing policy includes a first loss threshold, and the first machine learning model is used for managing a wireless communication service. The second communication data set is obtained by performing a processing operation on suspicious data in the first communication data set according to the first processing policy, and the suspicious data is wireless communication data whose loss value is greater than the first loss threshold.

TRAINING METHOD AND ELECTRONIC DEVICE FOR QUANTUM MACHINE LEARNING

Resumen de: US20260244971A1

This disclosure proposes a training method for quantum machine learning and an electronic device. The training method includes: configuring a quantum circuit to output probabilities of multiple qubits, where the quantum circuit comprises multiple gates with circuit parameters; mapping the qubits to multiple model parameters of a neural network, where multiple bases are calculated based on the qubits, and the quantity of the bases is greater than or equal to the quantity of the model parameters; inputting data into the neural network and calculating a loss based on the output of the neural network; and updating the circuit parameters in the quantum circuit according to the loss.

PREDICTING SUBJECTIVE RECOVERY FROM ACUTE EVENTS USING CONSUMER WEARABLES

Resumen de: US20260245733A1

0000 In an aspect, a method for predicting, for a subject, a recovery time from an acute or debilitating event is disclosed. The method may comprise (i) retrieving wearable sensor data from a first time period and a second time period. The first time period may be prior to the acute or debilitating event. The second time period may be after the acute or debilitating event. The method also may comprise (ii) determining the recovery time for the acute or debilitating event at least in part by processing said wearable sensor data from the first time period and the second time period with a trained machine learning algorithm.

MACHINE LEARNING MODEL PARAMETER TRANSFER

Resumen de: WO2026172331A1

Various aspects of the present disclosure relate to a node for wireless communication, comprising: at least one memory; and at least one processor coupled with the at least one memory and operable to cause the node to: obtain a machine learning (ML) model comprising of a set of model parameters; for each model parameter in the set of model parameters, determine a sensitivity of the model parameter, and provide error protection to the model parameter based on the sensitivity of the model parameter, the sensitivity of the model parameter indicating a degree of toleration for a value of the model parameter to vary without performance of the ML model degrading beyond a tolerance threshold value; and transmit the error protected model parameters to a further node.

PRIVACY PRESERVING WORKFLOW FOR REPRODUCIBLE MACHINE LEARNING TRAINING ITERATIONS

Resumen de: US20260244979A1

0000 Techniques for tracking machine learning model training iterations that preserve privacy, increase security, and reduce memory and other computing resources are disclosed herein. An example computer-implemented method comprises generating one or more hash values corresponding to raw data associated with a machine-learned model training process, the raw data comprising a first raw data point associated with a first key and the one or more hash values comprising a first hash value; storing the first hash values in a database in association with the first key; executing a first processing stage using the first raw data point to generate a processed data point; generating a hash value for the first processed data point; storing the second hash value in association with the first key; training a machine-learned model using processed data to generate a trained machine-learned model; and registering the trained machine-learned model.

MACHINE LEARNING-BASED AUTOMATED WELL LOG QUALITY CHECK AND RECONSTRUCTION

Resumen de: WO2026173837A1

A method for processing well log data includes obtaining input data including the well log data. The well log data may include a plurality of datasets. Each dataset of the plurality of datasets includes one or more log curves and is associated with a respective well of one or more wells. The method also includes harmonizing the plurality of datasets using a curated dictionary to produce harmonized datasets. The method further includes reconstructing the one or more log curves of each harmonized dataset of the harmonized datasets to produce reconstructed logs using a supervised machine-learning (ML) model. The method also includes generating normalized datasets based on the harmonized datasets and using log normalization. The method also includes generating an output based on the reconstructed logs and the normalized datasets.

LOCAL ARTIFICIAL INTELLIGENCE (AI)/MACHINE LEARNING (ML) SERVICE FOR INTERNET OF THINGS (IOT) DEVICES

Resumen de: US20260244994A1

0000 A device may include a processing device. The processing device may receive, at the device from an internet of things (IoT) device, one or more of an artificial intelligence (AI) task or a machine learning (ML) task. The processing device may receive, at the device from an IoT device, input data related to the one or more of the AI task or the ML task. The processing device may perform, at the device, the one or more of the AI task or the ML task using the input data to generate output data. The processing device may send, from the device to the IoT device, the output data.

SYSTEMS AND METHODS FOR IDENTITY GRAPH BASED FRAUD DETECTION

Resumen de: US20260245097A1

A method and apparatus for fraud detection during transactions using identity graphs are described. A method includes receiving, at a commerce platform system, a transaction from a user having initial transaction attributes and transaction data. The method also includes determining, by the commerce platform system, an identity associated with the user associated with additional transaction attributes not received with the transaction. Furthermore, the method includes accessing a feature set associated with the initial transaction attributes and the additional transaction attributes that includes machine learning (ML) model features for detecting transaction fraud. The method also includes performing, by the commerce platform system, a machine learning model analysis using the feature set and the transaction data to determine a likelihood that the transaction is fraudulent, and performing, by the commerce platforms system, the transaction when the likelihood that the transaction is fraudulent does not satisfy a transaction fraud threshold.

STABISMUS SURGERY OUTCOMES PREDICTION USING MACHINE LEARNING AND MULTIPLE PREOPERATIVE VARIABLES

Resumen de: WO2026172341A1

There is provided a system for generating a prediction for strabismus surgery, comprising: at least one processor executing a code for: receiving via a user interface and/or by reading data stored on a data storage device, input data including: a data- structure indicating a candidate surgery for strabismus surgery on at least one extraocular muscle of at least one eye of the subject, a preoperative deviation angle of the subject, and at least one parameter of the subject, and processing the input data by a machine learning model to generate a predicted change in the preoperative deviation angle in response to implementing the candidate surgery.

METHOD AND SYSTEM FOR GENERATING A MACHINE LEARNING MEDICAL DIFFERENTIAL DIAGNOSIS

Resumen de: WO2026171902A1

A computer implemented method for generating a machine learning medical differential diagnosis using a differential diagnosis coordination module that is in bidirectional communication with a plurality of agent modules that each perform a specific task. The method is performed by the differential diagnosis coordination module. The method comprises receiving patient profile data and performing a series of iterations that are carried out until a predetermined condition is satisfied. Each iteration comprises selecting an agent module from the plurality of agent modules, generating agent specific instructions that cause the selected agent module to perform its specific task according to the agent specific instructions, logging iteration attribute data that relates to attributes of a current iteration and that includes an output of the selected agent module generated according to the agent specific instructions, and updating the patient profile data based on the logged iteration attribute data. When the predetermined condition is satisfied, the method comprises outputting a medical differential diagnosis based on the logged iteration attribute data. The present disclosure can be used in a variety of applications including, but not limited to, several anticipated use cases in medical diagnostics/applications and in healthcare. The present disclosure can also help in patient/physician decision making and can be used with machine learning.

INFORMATION PROCESSING DEVICE, INFORMATION PROCESSING SYSTEM, AND INFORMATION PROCESSING METHOD

Resumen de: US20260244989A1

0000 There is provided a technology capable of making an extremely accurate and precise response to an inquiry including a cross-cutting issue like the one which requires a plurality of pieces of business knowledge. A device for managing a plurality of machine learning models respectively having specializations in specified fields, wherein the device includes at least a processor and a storage device; the processor: accepts information about specialization related to an inquiry from a user; selects a machine learning model candidate which contributes to generation of response content for the inquiry, from the plurality of machine learning models in the list based on the specialization related to the inquiry, the specialization information, and the key word; and decides a machine learning model to be used for the generation of the response content for the inquiry based on the selected machine learning model candidate.

MULTIMODAL DATA BASED GENERATION OF QUALITY ASSURED KNOWLEDGE GRAPH AND CONTEXTS FOR USER QUERIES

Resumen de: US20260245681A1

Gastrointestinal (GI) tract cancers represent a significant burden on global health, with their diagnosis often posing challenges due to overlapping symptoms and complex etiologies. Conventional methods are inaccurate in differentiating between various GI tract cancers and thus remain a formidable task for clinicians, often leading to delays in diagnosis and suboptimal management. Present disclosure provides a system and a method that receive multimodal data for generating a seed knowledge graph and patterns identification. Dynamic mapping is then performed using the identified patterns on the seed knowledge graph to obtain an updated seed knowledge graph using a deep learning model. The system then fuses the employs the updated seed knowledge graph with the multimodal data being processed to obtain multimodal patient profile. The system employs large language models (LLMs) to analyze patient data and generate insights and explainability, ensuring physicians understand the rationale behind diagnosis and treatment recommendations.

ARTIFICIAL INTELLIGENCE-BASED QUERY AND RESPONSE SYSTEMS AND METHODS

Resumen de: US20260244947A1

An AI-based computing system for responding in real-time to an inbound message includes a processor configured to: a) transmit, to a representative computing device, an AI model generated proposed response message responsive to a query message derived from the inbound message, b) receive, from the representative computing device, feedback associated with the AI model generated proposed response message, the feedback provided by a representative, d) create a historical record including the AI model generated proposed response message and the feedback, e) generate a training dataset including at least the created historical record, and g) using machine learning and/or artificial intelligence techniques, re-train the AI model using the training dataset.

META-MODEL FRAMEWORK FOR SIMULTANEOUS DUAL SCENARIO PREDICTION

Resumen de: US20260244997A1

Embodiments relate to technological systems and methods for determining return on investment (ROI) of a program (e.g., medical intervention), which may be used to retrospectively or prospectively evaluate the value that the program delivered or will deliver to a patient (e.g., a participant in a program). In some embodiments, a first set of training data associated with a patient population is retrieved. The training data may be classified as belonging to a patient treatment group including patients participating in a program or a patient control group including patients not participating in the program. A plurality of base machine learning models may be trained using a dual-training process to generate both counterfactual values and ROI predictions associated with patients. A global machine learning model may then be trained on the counterfactual values and ROI predictions output by the base machine learning models.

ARTIFICIAL INTELLIGENCE FOR PERSONALIZED GROWTH MODELS OF GEOGRAPHIC ATROPHY

Nº publicación: WO2026174075A1 20/08/2026

Solicitante:

UNIV COLORADO REGENTS [US]
THE REGENTS OF THE UNIVERSITY OF COLORADO, A BODY CORPORATE

Resumen de: WO2026174075A1

A computer-implemented method for modeling progression of geographic atrophy (GA) in a subject, comprising: receiving an image sequence of a subject's eye over time, the sequence comprising a baseline image; segmenting, via a first deep learning model, each image of the sequence to delineate a GA lesion boundary; computing a lesion area based on the GA lesion boundary; longitudinally registering, via a second deep learning model, the sequence to align each image to the baseline image; fitting a growth model of GA progression based on the computed lesion area of the sequence; estimating, via a Bayesian hierarchical framework, model parameters of the growth model; iteratively updating the growth model based on at least one subsequent image, thereby forming a digital twin of GA progression; and generating, based on the digital twin, an individualized forecast of GA lesion area, the individualized forecast comprising an associated uncertainty interval.