Alerta

DEVICE, SYSTEM AND METHOD FOR FEDERATED LEARNING USING RISK AUDITS

Absstract of: US20260025400A1

A computing device, that is configured to configure a global machine learning model, performs respective electronic risk audits of client devices configured to train respective local machine learning models that correspond to a global machine learning model. Based on respective electronic risk scores of one or more of the client devices, determined via the respective electronic risk audits, the computing device implements one or more parameter privacy adjustment methods on respective parameters received from the client devices prior to using the respective parameters to configure the global machine learning model, wherein respective client devices determined to have higher electronic risk scores have more of the parameter privacy adjustment methods applied than other respective client devices determined to have lower electronic risk scores. The computing device provides, to the client devices, the global machine learning model configured according to the respective parameters as adjusted.

Construction Knowledge Graph

Absstract of: US20260023889A1

An example computing platform is configured to (i) receive a data asset related to a construction project; (ii) determine, via a first machine-learning algorithm, at least one physical location within the construction project to which the received data asset is related; (iii) associate the received data asset with the determined physical location; (iv) based on the determined physical location, determine, via a second machine-learning algorithm, a respective relationship between the received data asset and one or more other data assets related to the construction project; and (v) add the received data asset to a construction knowledge graph as a node that is connected to one or more other respective nodes that represent the one or more other data assets.

EVALUATING ELECTRONIC SUBMISSIONS USING GENERATIVE ARTIFICIAL INTELLIGENCE

Absstract of: US20260023976A1

Aspects of the present disclosure relate to automated evaluation of electronic datasets. Embodiments include receiving one or more rules related to evaluation of electronic datasets. Embodiments further include generating, via an embedding model, embedding representations of the one or more rules. Embodiments further include receiving an electronic dataset. Embodiments further include identifying a rule that is applicable to the electronic dataset based on using a machine learning model configured to search the embedding representations of the one or more rules based on the electronic dataset. Embodiments further include evaluating, using the machine learning model or an additional machine learning model, the electronic dataset based on the identified rule. Embodiments further include using the machine learning model or the additional machine learning model to generate an evaluation summary for the electronic dataset based on determining that an item within the electronic dataset does not comply with the identified rule.

DEFECT-TRIGGERED MACHINE LEARNING-BASED TEST GENERATION AND CONTROL

Absstract of: US20260023678A1

An apparatus comprises at least one processing device configured to generate a first data structure by parsing a support ticket comprising information characterizing defects encountered while operating an information technology asset. The at least one processing device is also configured to process the first data structure utilizing a machine learning model to generate a second data structure specifying a given sequence of test steps for a given test scenario configured for testing of the defects. The at least one processing device is further configured to map the given sequence of test steps in the second data structure to respective application programming interface calls of a test automation framework, each of the application programming interface calls being associated with a functional code test unit of a test code database of the test automation framework, and to execute the given test scenario utilizing the mapped application programming interface calls.

Systems and Methods for Facial Recognition Training Dataset Adaptation with Limited User Feedback in Surveillance Systems

Absstract of: US20260023820A1

Various embodiments provide systems and methods for updating a training dataset so that the generated machine learning model can adapt to both short-term and long-term face variations including, for example, head pose, dressing, lighting conditions, and/or aging.

SYSTEMS AND METHODS FOR AUTONOMOUS TELEMETRY ORCHESTRATION

Absstract of: US20260024068A1

Systems, computer program products, and methods are described herein for autonomous telemetry orchestration. The present disclosure is configured to initiate and attempt transactions using IoT devices, generate unique session tokens, and verify session details against an orchestration engine by analyzing various parameters such as IP address, device ID, location, operating system, and mobile number. The system conducts a calculated score assessment and compares the score against a predefined threshold to determine transaction legitimacy. Transactions proceed if the score is below the threshold, otherwise, they are halted and alerts are issued. The system dynamically adjusts assessment models using machine learning algorithms based on historical data, employs blockchain technology for unique session tokens, and generates alerts via messaging services for suspicious activities.

REDUCING FALSE POSITIVES USING CUSTOMER FEEDBACK AND MACHINE LEARNING

Absstract of: US20260024101A1

A method of reducing a future amount of electronic fraud alerts includes receiving data detailing a financial transaction, inputting the data into a rules-based engine that generates an electronic fraud alert, transmitting the alert to a mobile device of a customer, and receiving from the mobile device customer feedback indicating that the alert was a false positive or otherwise erroneous. The method also includes inputting the data detailing the financial transaction into a machine learning program trained to (i) determine a reason why the false positive was generated, and (ii) then modify the rules-based engine to account for the reason why the false positive was generated, and to no longer generate electronic fraud alerts based upon (a) fact patterns similar to fact patterns of the financial transaction, or (b) data similar to the data detailing the financial transaction, to facilitate reducing an amount of future false positive fraud alerts.

METHOD AND APPARATUS FOR INTELLIGENT CLUSTERING OF CELLS AND NETWORK ENTITIES USING NETWORK INTELLIGENCE-AS-A-SERVICE

Absstract of: WO2026020169A1

Described are systems, apparatuses and methods for a machine learning k- means clustering in an Operations, Administration and Maintenance (OAM) module of a Radio Access Network to generate clusters of strongly-interfering cells together, while splitting apart weekly-interfering cells across different clusters.

VEHICLE TRAJECTORY TREE STRUCTURE INCLUDING LEARNED TRAJECTORIES

Absstract of: US20260021828A1

Techniques for generating a tree structure based on multiple machine-learned trajectories are described herein. A planning component (“ML system”) within a vehicle may receive and encode various types of sensor and/or vehicle data. The ML system can provide the encoded data as input to multiple machine-learning models (“ML models”), each of which may be trained to output a unique candidate trajectory for the vehicle follow. In some examples, each ML model may be trained to output a unique type of learned trajectory that causes the vehicle to perform a certain type of action. Using the learned candidate trajectories, the ML system may generate a tree structure that includes some or all of the candidate trajectories. The vehicle may determine a control trajectory based on the generation and traversal of the tree structure using a tree search algorithm, and may follow the control trajectory within the environment.

PREDICTION SELECTION FOR ITEM IDENTIFIERS USING EFFICIENT SELECTION ALGORITHM

Absstract of: WO2026019632A1

A smart system, such as a smart shopping cart system, uses an efficient selection algorithm to select an item identifier prediction for an item. The smart cart system uses a set of machine-learning models to generate identifier predictions based on images. To select an item identifier, the smart system applies an efficient selection algorithm to the predictions from the machine-learning models. An efficient selection algorithm is an algorithm that requires minimal computational resources to perform. For example, the efficient selection algorithm may be a simple majority algorithm that selects the identifier prediction generated by a majority of the models or a weighted voting algorithm where each model's vote is weighted by some metric. The smart system applies the efficient selection algorithm to select an item identifier prediction from the ones generated by the models. The smart system may display content related to the item associated with the item identifier prediction.

SYSTEM, METHOD, AND COMPUTER PROGRAM PRODUCT FOR INTEGRATED PROCESSING OF GENERATIVE AND INSTRUCTIVE PROMPTS IN MACHINE LEARNING MODELS

Nº publicación: WO2026019423A1 22/01/2026

Applicant:

VISA INT SERVICE ASS [US]
VISA INTERNATIONAL SERVICE ASSOCIATION

Absstract of: WO2026019423A1

Systems, methods, and computer program products are provided for integrated processing of generative and instructive prompts in machine learning models. An example system includes a processor configured to receive reference data, store a representation of the reference data, and receive a prompt. The processor is also configured to determine a first portion of the prompt associated with a generative prompt and a second portion of the prompt associated with an executable action. The processor is further configured to retrieve a subset of the representation and determine a generative output from a machine learning model based on the subset and the first portion of the prompt. The processor is further configured to generate content based on the generative output, determine an encoding of a plurality of action steps, and execute the executable action using a sequence-to-sequence decoder model and based on the content and the encoding.