Alerta

SYSTEM AND METHOD FOR MODEL ORCHESTRATION

Resumen de: US20260064302A1

A system for large-scale machine learning experiment execution, including: a platform configured to determine an experiment set from a run specification and schedule a run to one or more clusters; and a set of agents configured to receive the experiment set from the platform and facilitate individual experiment execution through a cluster orchestrator.

METHODS AND APPARATUS FOR TIME-SERIES FORECASTING USING DEEP LEARNING MODELS OF A DEEP BELIEF NETWORK WITH QUANTUM COMPUTING

Resumen de: US20260065109A1

An apparatus including a Deep Belief Network is configured to receive, via a processor, input data. The processor is caused to initialize, based on the input data, weights for a learning model of the DBN. The processor is further caused to generate, via the learning model, a representation of the input data. The weights, the input data, and the representation is to be transmitted to a quantum compute device. The processor is caused to receive sampled values from the quantum compute device using an optimization function associated with the quantum compute device. The processor is further caused to update, based on the sampled values, the weights to train the learning model to produce a trained learning model. The trained learning model is configured to generate an updated representation of the input data. The processor is further caused to generate, via a regression layer, output data based on the updated representation.

PROVIDING INTELLIGENT STORAGE LOCATION SUGGESTIONS

Resumen de: US20260065099A1

One or more embodiments of a content system provide machine-learned storage location recommendations for storing content items. Specifically, an online content management system can train a machine-learning model to identify a storage pattern from previously stored content items in a plurality of storage locations corresponding to a user account of a user. Training the machine-learning model includes training a plurality of classifiers for the plurality of storage locations. The online content management system uses the classifiers to determine whether a content item is similar to the content items in any of the storage locations, and based on the output of the classifiers, provides graphical elements indicating recommended storage locations within a graphical user interface. The user can select a graphical element to move the content item to the corresponding storage location.

MODEL UNINTERRUPTED SERVING AND EVOLUTION

Resumen de: US20260065166A1

Computer-implemented method and system for deployment of a first machine-learning model, and replacement, without service interruption, of a second machine-learning model in active on-line use, comprising: receiving, at a controller, a replacement request; in response to said replacement request, triggering the deployment of the first model and triggering the calculation of features to be used, collecting output data from the first model, fitting and inserting one or more calibration functions downstream from the first model, and routing inference requests to the first model instead of the second model; wherein the triggered deployment of the first model comprises preloading the first model into CPU or GPU memory, and making available the calculated features and the preloaded first model by CPU or GPU, respectively, before the inference requests are routed to the first model, thus enabling that no additional latency is added when traffic is rerouted to the first model.

METHOD FOR EXTRACTING ENTITIES AND RELATIONSHIPS FROM A CORPUS TO POPULATE A KNOWLEDGE GRAPH

Resumen de: AU2024336136A1

The present invention discloses a method for extracting entities and relationships from technical documents in a mostly automated way which achieves a more complete and accurate result in a reduced timeframe. Various deep learning models are trained using a corpus of the domain of interest annotated by experts and linguists. A vector graph model is also trained. Manual annotations and revisions are the minimum required to obtain automated models capable of automatically extracting entities and relationships from a corpus. Once trained, the models can be used on any corpus within the same knowledge domain.

MACHINE-LEARNING TECHNIQUES FOR PREDICTING UNOBSERVABLE OUTPUTS

Resumen de: AU2024322317A1

In some aspects, a computing system can generate and optimize a machine learning model to estimate an unobservable capacity of a target system or entity. The computing system can access training vectors which include training predictor variables, training performance indicators, and task quantities. A training performance indicator indicating performance outcome corresponding to the predictor variables and a task quantity associated with a task assigned to the target entity that leads to the training performance indicator. The machine learning model can be trained by performing adjustments of parameters of the machine learning model to minimize a loss function defined based on the training vectors. The trained machine learning model can be used to estimate the capacity of the target system or entity for handling tasks and be used in assigning tasks to the target entity according to the determined capacity.

VISUAL LOCATION OF AERIAL VEHICLES USING DYNAMIC ALEATORIC UNCERTAINTY

Resumen de: US20260063424A1

Techniques for localizing a vehicle in real time using dynamic uncertainty estimates are presented. The techniques include obtaining a terrain image captured by the vehicle; passing the terrain image to a trained evidential deep learning neural network subsystem, from which a dynamic uncertainty value and a first feature vector are obtained in real time; for each of a plurality of candidate terrain locations, comparing the first feature vector to a respective second feature vector representative of a candidate terrain location, from which a respective similarity score is obtained; for at least one of the plurality of candidate terrain locations, updating in real time, by a recursive Bayesian estimator, a respective location weight based on the dynamic uncertainty value and the respective similarity score; estimating, in real time, a location of the vehicle based on the plurality of location weights; and providing the location of the vehicle.

TRAINING A NEURAL DATABASE FOR ENTITY MATCHING

Resumen de: US20260065188A1

Certain aspects of the disclosure provide a method of training a neural database for entity matching. In examples, a method may include: extracting, from an electronic data repository, entity data related to a first entity that provides a good or a service; transforming the entity data into structured entity data configured to be processed by a machine learning model; processing the structured entity data with the machine learning model to generate metadata associated with the structured entity data; augmenting the structured entity data with the metadata associated with the structured entity data; and training the neural database based on the augmented structured entity data to predict one or more second entities that supply materials for the first entity and associated with the good or the service.

APPARATUS AND METHOD FOR EMERGENCY DISPATCH

Resumen de: US20260067398A1

An emergency data manager includes a mapping module that is operative to generate a map view in a cloud-based user interface provided to a public safety answering point (PSAP) by the emergency data manager. The map view displays location indicators for emergencies being handled by the PSAP. Machine learning trained logic is operatively coupled to the mapping module and is operative to correlate incoming emergency data and provide contextual data to PSAP dispatchers via the cloud-based user interface. The contextual data includes time, location, and event type. The machine learning trained logic may be further operative to provide a dispatch recommendation based on the contextual data, or based on contextual data and a set of dispatch rules. The machine learning trained logic may be further operative to provide a simulation of an experienced PSAP call taker or dispatcher.

DOUBLE TIER MACHINE LEARNING IN-SPACE HYBRID SIMULATIONS METHODS

Resumen de: EP4704091A1

A machine learning simulation method of determining a physical state of interaction between atoms from one or more physical properties of the atoms is disclosed. The method including dynamically evolving a first subset of atoms via a first machine learning model within a central high-fidelity region based on the one or more physical properties of the atoms. The method further includes dynamically evolving a second subset of the atoms via a second machine learning model with a remaining low-fidelity region based on the one or more physical properties of the atoms. The method also includes dynamically evolving a third subset of atoms located between the central high-fidelity region and the remaining low-fidelity region based on an interpolation of the first and second machine learning models to determine the physical state between the atoms.

TRAINING A NEURAL DATABASE FOR ENTITY MATCHING

Resumen de: EP4703968A1

Certain aspects of the disclosure provide a method of training a neural database for entity matching. In examples, a method may include: extracting, from an electronic data repository, entity data related to a first entity that provides a good or a service; transforming the entity data into structured entity data configured to be processed by a machine learning model; processing the structured entity data with the machine learning model to generate metadata associated with the structured entity data; augmenting the structured entity data with the metadata associated with the structured entity data; and training the neural database based on the augmented structured entity data to predict one or more second entities that supply materials for the first entity and associated with the good or the service.

SYSTEM AND METHOD FOR DETECTION AND BLOCKING OF FLASH CALLS

Resumen de: EP4703921A1

There is provided a method for training a machine learning model for identifying flash calls in a set of Call Detail Records, CDRs, the method comprising: receiving a set of CDRs and a set of call network data; creating a first training set, the first training set comprising a first subset of CDRs from the set of CDRs and a second subset of CDRs from the set of CDRs, wherein the first subset of CDRs comprises a plurality of CDRs known to represent flash calls, and the second subset of CDRs comprises a plurality of CDRs known to represent legitimate calls; determining one or more characteristic features in the first training set, the one or more characteristic features comprising a first characteristic feature associated with the first subset of CDRs and a second characteristic feature associated with the second subset of CDRs, wherein the first characteristic feature is different from the second characteristic feature.

METHOD AND SYSTEM FOR DEPLOYMENT AND REPLACEMENT OF A MACHINE-LEARNING MODEL IN ACTIVE ON-LINE USE WITHOUT SERVICE INTERRUPTION

Resumen de: EP4703874A1

Computer-implemented method and system for deployment of a first machine-learning model, and replacement, without service interruption, of a second machine-learning model in active on-line use, comprising: receiving, at a controller, a replacement request; in response to said replacement request, triggering the deployment of the first model and triggering the calculation of features to be used, collecting output data from the first model, fitting and inserting one or more calibration functions downstream from the first model, and routing inference requests to the first model instead of the second model; wherein the triggered deployment of the first model comprises preloading the first model into CPU or GPU memory, and making available the calculated features and the preloaded first model by CPU or GPU, respectively, before the inference requests are routed to the first model, thus enabling that no additional latency is added when traffic is rerouted to the first model.

ENHANCED FEATURE INDICATION FOR AI/ML POSITIONING FUNCTIONALITIES AND MODELS

Resumen de: WO2026043591A1

Aspects presented herein may enable a user equipment (UE) to indicate a correlation or a mapping between components of different feature groups (FGs) and differentiation of components in each FG to an artificial intelligence (AI) or machine learning (ML) (AI/ML) model/functionality level. In one aspect, a UE transmits, to a network entity, one or more first indications for a first positioning FG, where a number of the one or more first indications is indicative of a respective AI/ML model or functionality among a plurality of AI/ML models or functionalities associated with the first positioning FG and supported by the UE. The UE receives, from the network entity based on the one or more first indications, a second indication of at least one configuration for AI/ML positioning.

THREAT DETECTION PLATFORMS FOR DETECTING, CHARACTERIZING, AND REMEDIATING EMAIL-BASED THREATS IN REAL TIME

Resumen de: US20260058966A1

A method for behavior-based threat detection may include obtaining a first set of data corresponding to at least one of an employee or an enterprise associated with the employee. The method may include training a machine learning model for at least one of the employee or the enterprise associated with the employee by providing the first set of data to the machine learning model as training data, the machine learning model configured to identify deviations between behavioral traits of email communications and behavioral traits of the employee or the enterprise. The method may include receiving an email communication addressed to the employee. The method may include determining that the email communication represents a security risk by applying the machine learning model to the email communication. The method may include performing a remediation action on the email communication based on determining that the email communication represents a security risk.

WIRELESS DEVICE POWER OPTIMIZATION UTILIZING ARTIFICIAL INTELLIGENCE AND/OR MACHINE LEARNING

Resumen de: US20260059440A1

A method of reducing a power consumption of wireless communication circuitry of an edge device according to one embodiment includes determining a delivery traffic indication map (DTIM) interval of a wireless access point communicatively coupled to the edge device via the wireless communication circuitry of the edge device and adjusting a wake-up interval of the wireless communication circuitry of the edge device based on the DTIM interval to reduce the power consumption of the wireless communication circuitry of the edge device.

MANAGING COMPUTER RESOURCE USE WHEN CONDUCTING COMPUTER-IMPLEMENTED RISK ASSESSMENT

Resumen de: US20260056792A1

A method for managing computer resource use in computer-implemented risk assessment of a subject in respect of a context. Demographic subject data for the subject is passed to a first trained machine learning model trained on first context-specific historical risk outcomes correlated with historical demographic data corresponding to the demographic subject data. If a first threshold assessment from the first trained machine learning model is passed, the subject is approved. Responsive to failing the first threshold assessment, supplemental context-related subject data for the subject, in addition to the demographic subject data, is passed with the demographic subject data to a second trained machine learning model trained on second context-specific historical risk outcomes correlated with the historical demographic data and with historical context-related data corresponding to the supplemental context-related subject data. If a second threshold assessment from the second trained machine learning model is passed, the subject is approved.

HISTORICAL DATA RETENTION FOR AD MACHINE LEARNING MODELS

Resumen de: US20260057286A1

Described is a system for training a machine learning model by collecting a first dataset comprising ad impression data and ad conversion data over a first time period; training a first machine learning model using the first dataset; collecting a second dataset comprising ad impression data and ad conversion data over a second time period; selecting a subset of the first dataset; training a second machine learning model using a combined dataset of the subset of the first dataset and the second dataset to generate a second trained machine learning model configured to generate predicted ad conversion rates for new ads; applying a plurality of ads to the second trained machine learning model to receive individual predicted ad conversion rates for each of the plurality of ads; and ranking the ads based on the predicted ad conversion rates.

USER PROFILING USING CHAIN-OF-THOUGHT KNOWLEDGE GRAPHS FOR QUERYING A MACHINE LEARNING SYSTEM

Resumen de: US20260057254A1

Techniques are disclosed for a machine learning model, such as a large learning model (LLM), that incorporates a model of a chain of thought of a particular user when responding to a query from the user. In one example, a system generates a knowledge graph of a chain of thought of the user. The knowledge graph comprises nodes representing topics present within past queries by the user and edges representing a co-occurrence between the topics. The system determines, based on a topic present within a query from the user and the knowledge graph, a goal query comprising a goal topic. The system provides, to a machine learning model, the user to generate, by the machine learning model, a response. The machine learning model is constrained to include the goal topic of the goal query within the response. The system outputs, for display, the response to the query.

METHOD AND SYSTEM FOR AUTOMATED ERROR TRIAGING IN AN INDUSTRIAL PLANT

Resumen de: WO2026041226A1

The present invention relates to a method of performing error triaging in an industrial plant (106). The method involves receiving information (502) related to an incident in the industrial plant (106) and generating a knowledge graph (504) by analyzing a plurality of log files (108A-N). The knowledge graph (504) includes data on interdependencies among log events. The method further includes determining one or more nodes (506) of the knowledge graph (504) associated with a set of log events, generating a plurality of templates (508) for these log events using a first machine learning algorithm, and generating a summary report (516) for the incident by utilizing a large language model (514) to process the templates (508). This approach facilitates accurate and efficient identification, categorization, and reporting of errors within the industrial plant (106).

METHOD AND SYSTEM FOR CYBER THREAT DECEPTION

Resumen de: WO2026042071A1

The present invention relates to the technological field of cybersecurity, specifically to methods and systems for cyber threat deception using machine learning models. The claimed invention represents a system and method that provide an improvement of the technological field of cybersecurity by increasing efficiency and reliability of cyber threat deception, thereby enhancing the overall security posture of the network systems and improving the protection of genuine assets from potential breaches. The disclosed method and system addresses the challenges of the prior art by leveraging advanced machine learning techniques, to create a dynamic and adaptive deception approach.

MACHINE-LEARNING-BASED PREDICTION OF ADVERSE OUTCOMES USING DATA FROM IMPLANTABLE OR WEARABLE CARDIAC DEVICES

Resumen de: WO2026043905A1

A method for machine-learning-based prediction of adverse outcomes using measured time-varying values of physiological parameters generated from output of an implantable or wearable cardiac device includes receiving, as input to a trained machine learning model, measured time-varying values of physiological parameters generated from output of one or more embedded sensors of a wearable or implantable cardiac device worn by or implanted within an individual subject. The method further includes generating, as output from the trained machine learning model, a value indicating a personalized risk estimate of all-cause mortality or a composite event of all-cause mortality or heart failure hospitalization for the individual subject. The method further includes performing, based on the value indicating the personalized risk estimate of all-cause mortality or the composite event, an intervention for the individual subject.

GENERATIVE AI METHOD AND APPARATUS FOR ELECTRIC CIRCUIT LAYOUT DESIGN SYNTHESIS WITH PHYSICS-AUGMENTED MACHINE LEARNING MODELING

Resumen de: WO2026044026A1

An exemplary system and method for AI-based circuit layout design synthesis that employs a physics-augmented surrogate AI model and an optimizer to facilitate the design of low-loss impedance multi-layer radio-frequency circuit structures, other passive circuit structures, active circuit structures, among other circuit structures described herein. The synthesis tools can improve the design process of a high-performance circuit by orders of magnitude by using the AI surrogate model to predict the performance outcome of an intermediate design to generate a final optimzed synthetic design using only performance requirements as input to the synthesis process. The process can generate designs that also outperform conventional passive RF designs by providing the best structure from several optimized structures having been generated from different templates. The templates are preferably embedded in the latent training of the optimizer that can search the design space in an iterative manner without having to search the entire design space.

CATASTROPHIC WILDFIRE INDEX FOR FORECASTING UTILITY-CAUSED WILDFIRES

Resumen de: WO2026043960A1

A service determines a Hot Dry Windy (HDW) index for a vicinity around a utility component by inputting atmospheric conditions and weather conditions in the vicinity into a HDW model and receiving the HDW index as output from the HDW model. The service determines an Energy Release Component (ERC) percentile by inputting fuel loading and combustibility characteristics into an ERC model and receiving, as output from the ERC model, the ERC percentile. The service aggregates the HDW index and the ERC percentile into a modified HDW (mHDW) metric, inputs forecasted fire characteristics for the vicinity and the HDW index into a machine learning model, and receives as output from the machine learning model a likelihood of a fire growing to a threshold size. The service displays fire risk metric for the vicinity based on the likelihood of the fire growing to the threshold size.

METHOD AND SYSTEM FOR PROVIDING INTELLIGENT RESPONSE AGENT BASED ON SOPHISTICATED REASONING AND INFERENCE FUNCTION

Nº publicación: US20260056983A1 26/02/2026

Solicitante:

LG MAN DEVELOPMENT INSTITUTE CO LTD [KR]
LG MANAGEMENT DEVELOPMENT INSTITUTE CO., LTD

Resumen de: US20260056983A1

A method and system for providing an intelligent response agent based on a sophisticated reasoning and speculation function can generate and provide response data for queries related to specialized documents using a deep-learning neural network that implements a stepwise process for a sophisticated reasoning and speculation function.