Techniques for training data protection in an artificial intelligence model execution environment are disclosed. For example, a method comprises executing a first portion of an artificial intelligence model within a trusted execution area of an information processing system and a second portion of the artificial intelligence model within an untrusted execution area of the information processing system, wherein data at least one of obtained and processed in the first portion of the artificial intelligence model is inaccessible to the second portion of the artificial intelligence model. Data obtained in the trusted execution area may comprise one or more data samples in an encrypted form usable to train the artificial intelligence model.

Methods, systems, and techniques for kernel continuing learning. A dataset is obtained that corresponds to a classification task. Feature extraction is performed on the dataset using an artificial neural network. A kernel is constructed using features extracted during that feature extraction for use in performing the classification task. More particularly, during training, a coreset dataset corresponding to the classification task is saved; and during subsequent inference, the coreset dataset is retrieved and used to construct a task-specific kernel for classification.

Provided is a system for event forecasting using a graph-based machine-learning model that includes at least one processor programmed or configured to receive a dataset of data instances, where each data instance comprises a time series of data points, detect a plurality of motifs representing a plurality of events in the dataset of data instances using a matrix profile-based motif detection technique, generate a bipartite graph representation of the plurality of motifs in a time sequence, and generate a machine-learning model based on the bipartite graph representation of the plurality of motifs in the time sequence, where the machine-learning model is configured to provide an output and the output includes a prediction of whether an event will occur during a specified time interval. Methods and computer program products are also provided.

An online concierge system trains a user interaction model to predict a probability of a user performing an interaction after one or more content items are displayed to the user. This provides a measure of an effect of displaying content items to the user on the user performing one or more interactions. The user interaction model is trained from displaying content items to certain users of the online concierge system and withholding display of the content items to other users of the online concierge system. To train the user interaction model, the user interaction model is applied to labeled examples identifying a user and value based on interactions the user performed after one or more content items were displayed to the user and interactions the user performed when one or more content items were not used.

Certain aspects of the present disclosure provide techniques for recommending trip purposes to users of an application. Embodiments include receiving labeled travel data from the application running on a remote device including a plurality of trip purposes. Embodiments include building a topic model representing words associated with a plurality of topics. Embodiments include training a topic prediction model, using the plurality of topics and one or more features derived from each of the plurality of trip records, to output a topic based on an input trip record. Embodiments include training a purpose prediction model, using the topic model and the plurality of trip purposes, to output a trip purpose based on an input topic. The trip purpose may be recommended to a user via a user interface of the application running on the remote device.

This application provides an agent decision-making method and an apparatus, to improve decision-making performance of an agent. The method is applied to a communications system. The communications system includes at least two function modules. The at least two function modules include a first function module and a second function module, where the first function module is configured with a first agent, and the second function module is configured with a second agent. The method further includes the first agent obtaining related information of the second agent, and makes a decision on the first function module based on the related information of the second agent.

A method and a system for using a graph neural network framework to implement a link prediction in a multiplex network environment is provided. The method includes: identifying a plurality of layers of a multiplex network, each respective layer including a respective plurality of nodes; for each node included in at least a first layer, providing, by a structural node label and determining a common embedding across all of the plurality of layers and an individual embedding for each individual layer; using a k-nearest approach to select a subset of the plurality of layers for performing link prediction with respect to each layer based on the determined embeddings; and performing a link prediction by determining a respective feed-forward network with respect to each layer included in the selected subset.

In one aspect, an example methodology implementing the disclosed techniques includes receiving a corpus of historical order fulfillment data regarding a plurality of completed orders for one or more products, the historical order fulfillment data including an actual delivery time for each product in a completed order, and identifying, from the corpus of historical order fulfillment data, a plurality of features for a product, the plurality of features correlated with an actual delivery time for the product. The method also includes generating a training dataset using the identified plurality of features, the training dataset including a plurality of training samples, each training sample of the plurality of training samples corresponding to a product and including one or more identified features and the actual delivery time for the product. The method may include training the delivery time prediction module using the plurality of training samples.

The methods and systems are directed to computational approaches for training and using machine learning algorithms to predict the conditional marginal distributions of the position of agents at flexible evaluation horizons and can enables more efficient path planning. These methods model agent movement by training a deep neural network to predict the position of an agent through time. A neural ordinary differential equation (neural ODE) that represents this neural network can be used to determine the log-likelihood of the agent's position as it moves in time.

Methods, systems, and apparatus, including computer programs encoded on computer storage media, for performing persistent message passing using graph neural networks.