Alerta

HYBRID ARTIFICIAL INTELLIGENCE CLASSIFIER

Resumen de: WO2025181626A1

System, methods, apparatuses, and computer program products are disclosed for generating and using a hybrid artificial intelligence classifier for classifying input into one or more nodes of a taxonomy. Training data is received for at least a first portion of the taxonomy and used to train a supervised machine learning (ML) model to classify input into the first portion of the taxonomy having training data. A large language model (LLM) taxonomy is determined for at least a second portion of the taxonomy. The hybrid AI classifier classifies input based on a first classification obtained by providing the input to the supervised ML, and a second classification obtained by providing at least the input and the LLM taxonomy to a pre-trained LLM.

SYSTEMS AND METHODS FOR OPTIMIZING HYPERPARAMETERS OF MACHINE LEARNING MODELS USING REDUCTION ITERATION TECHNIQUES

Resumen de: EP4610891A1

The most fundamental task in ML models is to automate the setting of hyperparameters to optimize performance. Traditionally, in machine learning (ML) models hyperparameter optimization problem has been solved using brute-force techniques such as grid search, and the like. This strategy exponentially increases computation costs and memory overhead. Considering the complexity and variety of the ML models there still remains practical difficulties of selecting right combinations of hyperparameters to maximize performance of the ML models. Embodiments of the present disclosure provide systems and methods for hyperparameters optimization in machine learning models and to effectively reduce the hyperparameter search dimensions and identify the important hyperparameter dimensions that are high variable to identify the best hyperparameter thereby saving the computing energy of machine learning process and eliminate categorical dimensions by using a combination of reduction-iteration techniques.

SYSTEMS AND METHODS FOR AUTHENTICATING A RESOURCE SYSTEM

Resumen de: US2025274452A1

Systems and methods are disclosed for determining authenticity of a resource system. The method includes receiving a dataset that includes a first subset and a second subset associated with a first resource system; down-sampling the first subset but not the second subset; generating a first feature for a machine learning model based on the down-sampled first subset; generating a second feature for the machine learning model based on the second subset; generating, via input of at least one of the first feature or the second feature into the machine learning model that is trained to output a fraudulent measure, one or more data objects indicative of validating the fraudulent measure; and initiating performance of one or more prediction-based actions in response to the generating.

DETECTING RANSOMWARE IN MONITORED DATA

Resumen de: US2025272394A1

An information management system includes one or more client computing devices in communication with a storage manager and a secondary storage computing device. The storage manager manages the primary data of the one or more client computing devices and the secondary storage computing device manages secondary copies of the primary data of the one or more client computing devices. Each client computing device may be configured with a ransomware protection monitoring application that monitors for changes in their primary data. The ransomware protection monitoring application may input the changes detected in the primary data into a machine-learning classifier, where the classifier generates an output indicative of whether a client computing device has been affected by malware and/or ransomware. Using a virtual machine host, a virtual machine copy of an affected client computing device may be instantiated using a secondary copy of primary data of the affected client computing device.

INFORMATION PROCESSING SYSTEM, INFORMATION PROCESSING METHOD, RECORDING MEDIUM

Resumen de: US2025272457A1

An information processing method performed by an information processing unit includes obtaining descriptors related to a material, converting the obtained descriptors into correlativity-reduced descriptors in which correlativity therebetween is reduced by using conversion coefficients, calculating importance of the correlativity-reduced descriptors, identifying important descriptors from the descriptors on a basis of the conversion coefficients used for conversion into the correlativity-reduced descriptors having high importance, generating a learned model by machine learning using the important descriptors and a characteristic value of the material as learning data, and estimating a characteristic value of a different material different from the material by using the learned model.

SYSTEM AND METHOD FOR FEEDBACK-DRIVEN AUTOMATED DRUG DISCOVERY

Resumen de: US2025272582A1

A system and method for feedback-driven automated drug discovery which combines machine learning algorithms with automated research facilities and equipment to make the process of drug discovery more data driven and less reliant on intuitive decision-making by experts. In an embodiment, the system comprises automated research equipment configured to perform automated assays of chemical compounds, a data platform comprising drug databases and an analysis engine, a bioactivity and de novo modules operating on the data platform, and a retrosynthesis system operating on the drug discovery platform, all configured in a feedback loop that drives drug discovery by using the outcome of assays performed on the automated research equipment to feed the bioactivity module and retrosynthesis systems, which identify new molecules for testing by the automated research equipment.

SYSTEMS AND METHODS FOR GENERATING CARBON EFFICIENT CODE

Resumen de: US2025272067A1

In some aspects, the techniques described herein relate to a method including: determining an optimal code block from a plurality of code blocks, wherein the plurality of code blocks each solve a predetermined problem, and wherein the optimal code block optimizes a coding efficiency; pairing the optimal code block with suboptimal code blocks from the plurality of code blocks, wherein the pairing generates suboptimal-to-optimal code pairs; training a machine learning model with the suboptimal-to-optimal code pairs, wherein the training fine-tunes the machine learning model for inferring edits to input code blocks, and wherein the edits to the input code blocks optimize the input code blocks in terms of the coding efficiency; receiving, at the machine learning model, an input code block; inferring, by the machine learning model, edits to the input code block; and outputting, by the machine learning model, an optimized code block based on the input code block.

ADVANCED SYSTEMS AND METHODS FOR MULTIMODAL AI: GENERATIVE MULTIMODAL LARGE LANGUAGE AND DEEP LEARNING MODELS WITH APPLICATIONS ACROSS DIVERSE DOMAINS

Resumen de: US2025272534A1

Systems and methods are provided for improving generative artificial intelligence (AI). Systems and methods can integrate more reliable data sources and enhance generative AI training and inference processes for complex tasks. The integration of real-time data and expert input can be included as crucial steps in aligning AI outputs with improved accuracy. Similarly, fine-tuning methodologies and augmentation algorithms can be used to focus on minimizing the occurrence of fabricated content, thereby significantly increasing the chances that the information generated is both current and credible.

AUTOMATED MULTIVARIATE SYSTEM PERFORMANCE ANALYSIS

Resumen de: US2025272566A1

The embodiments described herein generally relate to automated performance analysis of a system. Embodiments include receiving parameter values for a plurality of parameters captured during a time period. Embodiments include providing inputs based on the data set to a supervised machine learning model configured to determine significant parameters with respect to a target variable. Embodiments include receiving, from the supervised machine learning model in response to the inputs, an indication of two or more significant parameters from the plurality of parameters with respect to the target variable. Embodiments include generating a multivariate cluster for the target variable based on the two or more significant parameters and determining an anomalous state of the system with respect to the target variable based on the multivariate cluster for the target variable and data captured after the time period.

ON-SITE UPDATING OF MACHINE LEARNING MODELS AND MACHINE LEARNING MODELS INCORPORATING HARDWARE AND RUNTIME ATTRIBUTES

Resumen de: US2025272599A1

Updating machine learning models with user data includes executing, by a data processing system, a container including a first machine learning (ML) model, training data for the first ML model, and a library of machine learning functions. The data processing system executes one or more of the machine learning functions of the library. The one or more of the machine learning functions are configured to build a second ML model trained, at least in part, on user training data and to compare accuracy of the first ML model with accuracy of the second ML model. An ML model also may be trained to predict compilation time for circuit designs using training data that includes circuit design features, hardware features of a data processing system, and runtime features from the data processing system.

TECHNIQUES FOR DESIGNING SYSTEMS WITH MULTI-OBJECTIVE BAYESIAN OPTIMIZATION

Resumen de: US2025272586A1

One embodiment of a method for designing a system includes processing historical data associated with zero or more previous designs of the system using a trained machine learning model to predict a plurality of rewards for a plurality of designs of the system that are associated with different combinations of parameter values, and selecting, from the plurality of designs of the system, a first design of the system that is associated with a highest reward included in the plurality of rewards.

CONDITIONAL CONFORMAL PREDICTION INTERVALS

Resumen de: US2025272578A1

An apparatus for computing a conditional conformal prediction interval for a machine learning point prediction regression model and calibration point predictions forming a distribution of an error around the point prediction regression model in an input space. The apparatus includes a conformal regions circuit configured to compute a quantile regression of the error to compute an approximation of a quantile of the error. The conformal regions circuit is further configured to identify a set of regions in the input space where the distribution within each region in the set of regions is interpretably constant. In one embodiment, the apparatus also includes a conformal prediction circuit configured to compute the conditional conformal prediction interval for the point prediction regression model conditioned on the identified set of regions and the corresponding computed quantile of the error for each region in the set of regions.

Machine Learning Framework for Human Data Detection

Resumen de: US2025272543A1

Methods, systems, and apparatuses are described herein for detecting human data. A machine learning model may be trained to predict, based on an input vector representing a column name, whether the column name corresponds to human data. A computing device may process, using a fuzzy string-matching algorithm, a column name to determine if the column name corresponds to one or more known human data categories. If the fuzzy string-matching algorithm does not find a match, an expanded column name might be generated by mapping the column name to a vector space using a transformer model. That expanded column name may be provided to one or more input nodes of the trained machine learning model and, based on output from the trained machine learning model, the computing device may store metadata, associated with the first column, that indicates whether the first column comprises human data.

MACHINE LEARNING MODEL TRAINING ON RISK PREDICTION USING GRAPH KNOWLEDGE DISTILLATION

Resumen de: US2025272552A1

Various embodiments described herein support or provide operations including identifying a machine-learning (ML) model associated with an omni-view knowledge graph; generating an embedding vector that represents the omni-view knowledge graph; identifying a ML model associated with a temporal-view knowledge graph; generating an embedding vector that represents the temporal-view knowledge graph; and training a ML model based on the generated embedding vectors.

GENERATION OF USER-SPECIFIC ELECTRONIC PROMPTS FOR A COMPUTING-BASED PROCESS

Resumen de: US2025272569A1

An intelligent workflow process is provided to facilitate generating electronic prompt(s) for a user of a computing system to provide customized assistance to the user in carrying out a computing-based process. Generating the prompt(s) includes identifying, by an artificial intelligence agent with reference to user data, the computing-based process initiated by the user, where the user data includes historical data relevant to the computing-based process. In addition, the generating includes determining, by the artificial intelligence agent using a machine learning model and the user data, one or more typical actions of the user relevant to the computing-based process, and producing, based on the one or more typical actions of the user relevant to the computing-based process, the electronic prompt(s) for the user. In addition, the electronic prompt(s) are provided, to the user's computing system to facilitate the customized assistance to the user in carrying out the computing-based process.

SYSTEM AND METHOD FOR EVALUATING AN UNSUPERVISED CLUSTERING MACHINE LEARNING (ML) MODEL

Resumen de: US2025272603A1

Disclosed herein is a method for evaluating an unsupervised clustering machine learning (ML) model. The method includes generating a set of model clusters via the unsupervised clustering ML model. Further, the method includes comparing a set of test set clusters and the set of model clusters. Further, the method includes categorizing each of the set of model clusters into an assessment group based on the comparison. The categorized assessment group is at least one of a match group, a correct group, a partial group, and an incorrect group. Furthermore, the method includes assigning a similarity value to each of the set of model clusters based on the categorized assessment group. Furthermore, the method includes determining a total similarity value based on combining the assigned similarity value of each of the set of model clusters, such that the total similarity value indicates evaluation of the unsupervised clustering ML model.

MACHINE LEARNING MODEL BASED ARCHITECTURE FOR QUERY SERVICES

Resumen de: US2025272604A1

Technical solutions include an ML based multi-model architecture to generate responses to enterprise employee queries. A processor can receive a query on a topic and identify ML models for a plurality of domains, trained using texts on a respective domain of the plurality of domains for each respective ML model and covering a plurality of topics corresponding geographic areas. The processor can select, using a first portion of the query and a classification model trained to classify the ML models according to the topics, a first ML model trained on a domain associated with the topic of the query. The processor can generate, using a second portion of the query corresponding to a geographic area of the geographic areas and the first ML model, a response to the query and provide the response.

QUANTUM COMPUTING AND AI BIDIRECTIONAL MONITORING WITH QUANTUM COMPUTING AS A FLEXIBLE GUARDRAIL TO AI

Resumen de: US2025272591A1

Systems and methods for artificial intelligence (“AI”) bidirectional monitoring with a quantum-computing-powered system as a flexible guardrail to AI are provided. The systems and methods may include a quantum processor and a classical processor. The systems and methods may include requesting data elements pertaining to boundaries. The systems and methods may include controlling boundary rules and creating classical boundary rules via a classical processor. The systems and methods may include interfacing classical boundary rules with a quantum processor. The systems and methods may include running Grover's conversions in parallel over the boundary rules. The systems and methods may include pulling dynamic market data through a legacy transformation platform including dynamically derived data values and a machine learning model (“MLM”) thereby monitoring and controlling an AI and machine learning (“ML”) processor.

HYBRID ARTIFICIAL INTELLIGENCE CLASSIFIER

Resumen de: US2025272608A1

System, methods, apparatuses, and computer program products are disclosed for generating and using a hybrid artificial intelligence classifier for classifying input into one or more nodes of a taxonomy. Training data is received for at least a first portion of the taxonomy and used to train a supervised machine learning (ML) model to classify input into the first portion of the taxonomy having training data. A large language model (LLM) taxonomy is determined for at least a second portion of the taxonomy. The hybrid AI classifier classifies input based on a first classification obtained by providing the input to the supervised ML, and a second classification obtained by providing at least the input and the LLM taxonomy to a pre-trained LLM.

UTILIZING A TRACTABILITY MACHINE LEARNING MODEL TO GENERATE TRACTABILITY SCORES FOR A MULTI-DOMAIN MACHINE LEARNING MODEL FOR IMPROVED MACHINE LEARNING PREDICTIONS

Resumen de: US2025272607A1

The present disclosure relates to systems, non-transitory computer-readable media, and methods that utilize a multi-domain tractability machine learning model to generate tractability scores for a multi-domain machine learning model and further generate improved bioactivity predictions. Indeed, in one or more implementations, the disclosed systems generate a predicted match score between a target protein and a target compound using a compound-protein interaction machine learning model. For instance, the disclosed systems generate a protein-model tractability score that indicates a measure of accuracy of the compound-protein interaction machine learning model relative to the target protein. Moreover, in some instances, the disclosed systems utilize the protein-model tractability score by providing the protein-model tractability score in conjunction with the predicted match score or the target protein or the disclosed systems generate a bioactivity prediction from the predicted match score and the protein-model tractability score.

EXTENSIBLE MACHINE LEARNING POWERED BEHAVIORAL FRAMEWORK FOR RISK COVERAGE

Resumen de: US2025272617A1

Some aspects of the present disclosure relate to systems, methods and computer readable media for outputting alerts based on potential violations of predetermined standards of behavior. In one example implementation, a computer implemented method includes: training a natural language-based machine learning model to detect at least one risk of a violation condition in an electronic communication between persons, wherein the violation condition is a potential violation of a first predetermined standard of behavior; receiving a lexicon, wherein the lexicon comprises topic data; receiving connection data representing a relationship between the trained machine learning model and the lexicon; detecting, using the trained machine learning model, the lexicon, and the connection data, a potential violation of a second predetermined standard of behavior; and outputting for display an alert indicating the potential violation of the second predetermined standard of behavior.

Machine Learning Model for Identifying Expertise within an Organization

Resumen de: US2025272627A1

Systems and methods for providing directory support in an organization are described. A method, according to one implementation, includes gathering digital data from multiple sources within an organization. The method also includes indexing the digital data in a table that includes at least a first column including names of a plurality of members of the organization and a second column including keywords attributed to the plurality of members. Also, the method includes training a Machine Learning (ML) model by data crawling through the digital data, assigning weights to the keywords, and storing the weights in a third column in the table. In response to receiving an inquiry from a user, the ML model is configured during inference to use information in the table to provide an output to the user identifying a member in the organization who demonstrates expertise on a specific topic.

ADVANCED SYSTEMS AND METHODS FOR MULTI-MODAL AI: GENERATIVE MULTI-MODAL LARGE LANGUAGE AND DEEP LEARNING MODELS WITH APPLICATIONS ACROSS DIVERSE DOMAINS

Resumen de: WO2025178877A1

Systems and methods are provided for improving generative artificial intelligence (Al). Systems and methods can integrate more reliable data sources and enhance generative Al training and inference processes for complex tasks. The integration of real-time data and expert input can be included as crucial steps in aligning Al outputs with improved accuracy. Similarly, fine-tuning methodologies and augmentation algorithms can be used to focus on minimizing the occurrence of fabricated content, thereby significantly increasing the chances that the information generated is both current and credible.

Methods and Systems to Validate Physiologic Waveform Reliability and Uses Thereof

Resumen de: US2025269112A1

Methods and systems to validated physiologic waveform reliability and uses thereof are provided. A number of embodiments describe methods to validate waveform reliability, including blood pressure waveforms, electrocardiogram waveforms, and/or any other physiological measurement producing a continuous waveform. Certain embodiments output reliability measurements to closed loop systems that can control infusion rates of cardioactive drugs or other fluids in order to regulate blood pressure, cardiac rate, cardiac contractility, and/or vasomotor tone. Further embodiments allow for waveform evaluators to validate waveform reliability based on at least one waveform feature using data collected from clinical monitors using machine learning algorithms.

TEST ASSIST APPARATUS, TEST ASSIST METHOD, AND RECORDING MEDIUM

Nº publicación: US2025273307A1 28/08/2025

Solicitante:

NEC CORP [JP]
NEC Corporation

Resumen de: US2025273307A1

This test assist apparatus includes: an acquiring section for acquiring utterance information which indicates the content of an utterance of a patient during a test performed on the patient, state information regarding the feelings of the patient during the test, and basic information regarding the test; an interruption predicting section for predicting a probability of interruption of the test performed on the patient, from the utterance information, the state information, and the basic information, with use of a prediction model for predicting a probability of interruption of a test, the prediction model being generated by machine learning in which samples of past test interruptions are used as training data; and an outputting section for outputting the probability of interruption.