Alerta

SYSTEMS AND METHODS OF OPTIMIZING RESOURCE ALLOCATION USING MACHINE LEARNING AND PREDICTIVE CONTROL

Resumen de: US2025245050A1

A computer system includes a transceiver that receives over a data communications network different types of input data from multiple source nodes and a processing system that defines for each of multiple data categories, a set of groups of data objects for the data category based on the different types of input data. Predictive machine learning model(s) predict a selection score for each group of data objects in the set of groups of data objects for the data category for a predetermined time period. Control machine learning model(s) determine how many data objects are permitted for each group of data objects based on the selection score. Decision-making machine learning model(s) prioritize the permitted data objects based on one or more predetermined priority criteria. Subsequent activities of the computer system are monitored to calculate performance metrics for each group of data objects and for data objects actually selected during the predetermined time period. Predictive machine learning model(s) and decision-making machine learning model(s) are adjusted based on the performance metrics to improve respective performance(s).

Load Balancing For Mixture Of Experts Machine Learning

Resumen de: US2025245553A1

Aspects of the disclosure are directed to improving load balancing for serving mixture of experts (MoE) machine learning models. Load balancing is improved by providing memory dies increased access to computing dies through a 2.5D configuration and/or an optical configuration. Load balancing is further improved through a synchronization mechanism that determines an optical split of batches of data across the computing die based on a received MoE request to process the batches of data. The 2.5D configuration and/or optical configuration as well as the synchronization mechanism can improve usage of the computing die and reduce the amount of memory dies required to serve the MoE models, resulting in less consumption of power and lower latencies and complexity in alignment associated with remotely accessing memory.

On-Device Machine Learning Platform

Resumen de: US2025245533A1

The present disclosure provides systems and methods for on-device machine learning. In particular, the present disclosure is directed to an on-device machine learning platform and associated techniques that enable on-device prediction, training, example collection, and/or other machine learning tasks or functionality. The on-device machine learning platform can include a context provider that securely injects context features into collected training examples and/or client-provided input data used to generate predictions/inferences. Thus, the on-device machine learning platform can enable centralized training example collection, model training, and usage of machine-learned models as a service to applications or other clients.

CLINICAL DATA ANALYSIS

Resumen de: WO2025157774A1

Computer-implemented methods of providing a clinical predictor tool are described, comprising: obtaining training data comprising, for each of a plurality of patients, values for a plurality of clinical variables comprising a variable indicative of a diagnosis or prognosis and one or more further clinical variables; and training a clinical predictor model to predict the variable indicative of a diagnosis or prognosis using said training data, wherein obtaining the training data comprises obtaining synthetic clinical data comprising values for a plurality of clinical variables for one or more patients by obtaining a directed acyclic graph (DAG) edges corresponding to conditional dependence relationships inferred from real clinical data comprising values for the plurality of clinical variables for a plurality of patients, and obtaining values for each node of the DAG using a machine learning model and multivariate conditional probability table. Computer-implemented methods of obtaining synthetic clinical data are also described.

Using Optical Remote Sensors And Machine Learning Models To Predict Agronomic Field Property Data

Resumen de: US2025245532A1

In some embodiments, a computer-implemented method for predicting agronomic field property data for one or more agronomic fields using a trained machine learning model is disclosed. The method comprises receiving, at an agricultural intelligence computer system, agronomic training data; training a machine learning model, at the agricultural intelligence computer system, using the agronomic training data; in response to receiving a request from a client computing device for agronomic field property data for one or more agronomic fields, automatically predicting the agronomic field property data for the one or more agronomic fields using the machine learning model configured to predict agronomic field property data; based on the agronomic field property data, automatically generating a first graphical representation; and causing to display the first graphical representation on the client computing device.

ADAPTIVE LENGTH SPECULATIVE DECODING IN AUTOREGRESSIVE GENERATIVE ARTIFICIAL INTELLIGENCE MODELS

Resumen de: US2025245530A1

Certain aspects of the present disclosure provide techniques and apparatus for generating a response to a query input in a generative artificial intelligence model using variable draft length. An example method generally includes determining (e.g., measuring or accessing) one or more operational properties of a device on which inferencing operations using a machine learning model are performed. A first draft set of tokens is generated using the machine learning model. A number of tokens included in the first draft set of tokens is based on the one or more operational properties of the device and a defined scheduling function for the machine learning model. The first draft set of tokens are output for verification.

LOAD BALANCING FOR MIXTURE OF EXPERTS MACHINE LEARNING

Resumen de: EP4592895A1

Aspects of the disclosure are directed to improving load balancing for serving mixture of experts (MoE) machine learning models. Load balancing is improved by providing memory dies increased access to computing dies through a 2.5D configuration and/or an optical configuration. Load balancing is further improved through a synchronization mechanism that determines an optical split of batches of data across the computing die based on a received MoE request to process the batches of data. The 2.5D configuration and/or optical configuration as well as the synchronization mechanism can improve usage of the computing die and reduce the amount of memory dies required to serve the MoE models, resulting in less consumption of power and lower latencies and complexity in alignment associated with remotely accessing memory.

TRAINING A MACHINE-LEARNING MODEL FOR CONSTRAINT-COMPLIANCE PREDICTION USING AN ACTION-BASED LOSS FUNCTION

Resumen de: WO2024081965A1

An online system trains a constraint prediction machine-learning model using an action-based loss function. The action-based loss function computes a weighted sum of (1) an accuracy of the constraint prediction model in predicting whether a user's interaction complies with a set of constraints and (2) an action score representing a number of actions taken by users to determine whether the user's action complies with the set of constraints. The online system may apply this trained constraint prediction model to future interaction data received from third-party systems to predict whether user interactions with those third-party systems comply with the set of constraints.

FORWARD-FORWARD TRAINING FOR MACHINE LEARNING

Resumen de: WO2024112887A1

Example implementations provide a computer-implemented method for training a machine-learned model, the method comprising: processing, using a layer of the machine-learned model, positive input data in a first forward pass; updating one or more weights of the layer to adjust, in a first direction, a goodness metric of the layer for the first forward pass; processing, using the layer, negative input data in a second forward pass; and updating the one or more weights to adjust, in a second direction, the goodness metric of the layer for the second forward pass.

MODELLING CAUSATION IN MACHINE LEARNING

Resumen de: US2024104338A1

A method comprising: sampling a temporal causal graph from a temporal graph distribution specifying probabilities of directed causal edges between different variables of a feature vector at a present time step, and from one variable at a preceding time step to another variables at the present time step. Based on this there are identified: a present parent which is a cause of the selected variable in the present time step, and a preceding parent which is a cause of the selected variable from the preceding time step. The method then comprises: inputting a value of each identified present and preceding parent into a respective encoder, resulting in a respective embedding of each of the present and preceding parents; combining the embeddings of the present and preceding parents, resulting in a combined embedding; inputting the combined embedding into a decoder, resulting in a reconstructed value of the selected variable.

MODELLING CAUSATION IN MACHINE LEARNING

Nº publicación: EP4591216A1 30/07/2025

Solicitante:

MICROSOFT TECHNOLOGY LICENSING LLC [US]
Microsoft Technology Licensing, LLC

Resumen de: US2024104370A1

A method comprising: sampling a first causal graph from a first graph distribution modelling causation between variables in a feature vector, and sampling a second causal graph from a second graph distribution modelling presence of possible confounders, a confounder being an unobserved cause of both of two variables. The method further comprises: identifying a parent variable which is a cause of a selected variable according to the first causal graph, and which together with the selected variable forms a confounded pair having a respective confounder being a cause of both according to the second causal graph. A machine learning model encodes the parent to give a first embedding, and encodes information on the confounded pair give a second embedding. The embeddings are combined and then decoded to give a reconstructed value. This mechanism may be used in training the model or in treatment effect estimation.