Abstract: Embodiments of the present invention provide a machine learning model training method, including: obtaining target task training data and N categories of support task training data; inputting the target task training data and the N categories of support task training data into a memory model to obtain target task training feature data and N categories of support task training feature data; training the target task model based on the target task training feature data and obtaining a first loss of the target task model, and separately training respectively corresponding support task models based on the N categories of support task training feature data and obtaining respective second losses of the N support task models; and updating the memory model, the target task model, and the N support task models based on the first loss and the respective second losses of the N support task models.

Abstract: A learning device is configured to perform learning of a decision tree. The learning device includes a gradient output unit and a branch score calculator. The gradient output unit is configured to output a cumulative sum of gradient information corresponding to each value of a feature amount of learning data. The branch score calculator is configured to calculate a branch score used for determining a branch condition for a node of the decision tree, from the cumulative sum without using a dividing circuit.

Abstract: A method of updating a first neural network is disclosed. The method includes providing a computer system with a computer-readable memory that stores specific computer-executable instructions for the first neural network and a second neural network separate from the first neural network. The method also includes providing one or more processors in communication with the computer-readable memory. The one or more processors are programmed by the computer-executable instructions to at least process a first data with the first neural network, process a second data with the second neural network, update a weight in a node of the second neural network by a delta amount as a function of the processing of the second data with the second neural network, and update a weight in a node of the first neural network as a function of the delta amount. A computer system for updating a first neural network is also disclosed. Other features of the preferred embodiments are also disclosed.

Abstract: A system receives information associated with an interaction with an individual in a context. Then, the system analyzes the information to extract features associated with one or more attributes of the individual. Moreover, the system generates, based at least in part on the extracted features, a group of behavioral agents in a multi-layer hierarchy that automatically mimics the one or more attributes. Next, the system calculates one or more performance metrics associated with the group of behavioral agents and the one or more attributes. Furthermore, the system determines, based at least in part on the one or more performance metrics, one or more deficiencies in the extracted features. Additionally, the system selectively acquires second information associated with additional interaction with the individual in the context based at least in part on the one or more deficiencies to at least in part correct for the one or more deficiencies.

Abstract: An equilibrium computation acceleration method and system for a sparse recurrent neural network determine scheduling information based on an arbitration result of sparsity of a weight matrix, select a computation submodule having operating voltage and operating frequency that match the scheduling information or a computation submodule having operating voltage and operating frequency that are adjusted to match the scheduling information, and use the selected computation submodule to perform a zero-hop operation and a multiply-add operation in sequence, to accelerate equilibrium computation. The equilibrium computation acceleration system includes a data transmission module, an equilibrium computation scheduling module with a plurality of independent built-in computation submodules, and a voltage-adjustable equilibrium computation module. An error monitor is configured to achieve dynamic voltage adjustment.

Abstract: A resistive processing unit (RPU) that includes a pair of transistors connected in series providing an update function for a weight of a training methodology to the RPU, and a read transistor for reading the weight of the training methodology. In some embodiments, the resistive processing unit (RPU) further includes a capacitor connecting a gate of the read transistor to the air of transistors providing the update function for the resistive processing unit (RPU). The capacitor stores said weight of training methodology for the RPU.

Abstract: Effectively training machine learning systems with incomplete/partial labels is a practical, technical problem that solutions described herein attempt to overcome. In particular, an approach to modify loss functions on a proportionality basis is noted in some embodiments. In other embodiments, a graph neural network is provided to help identify correlations/causations as between categories. In another set of embodiments, a prediction approach is described to, based on originally provided labels, predict labels for unlabelled training samples such that the proportion of labelled labels relative to all labels is increased.

Abstract: Disclosed are a data processing method, a device and a medium. The method includes: acquiring first feature data and a source identification of data to be processed; determining a first unshared hidden unit, corresponding to the source identification, in an autoencoder, wherein the autoencoder includes a plurality of first unshared hidden units that do not share a parameter with each other; inputting the first feature data into the determined first unshared hidden unit, to perform noise cancellation, and outputting second feature data meeting a set standard; inputting the second feature data into a first shared hidden unit of the autoencoder to map the second feature data to a set feature space through the first shared hidden unit, and outputting mapping data; and inputting the mapping data into a shared feature layer of the autoencoder, and outputting common feature data in the first feature data, extracted by the shared feature layer.

Abstract: A single neural network model can be used by each computing engine (CE) in a neural network processor to perform convolution operations in parallel for one or more stacks of convolutional layers. An input feature map can be divided into N chunks to be processed by N CEs, respectively. Each CE can process a last portion of a respective chunk to generate respective shared states to be used by a subsequent CE. A first CE uses pre-computed states to generate a first portion of an output feature map, while other CEs use shared states computed by a preceding CE to generate respective portions of the output feature map.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for augmenting machine learning language models using search engine results. One of the methods includes obtaining question data representing a question; generating, from the question data, a search engine query for a search engine; obtaining a plurality of documents identified by the search engine in response to processing the search engine query; generating, from the plurality of documents, a plurality of conditioning inputs each representing at least a portion of one or more of the obtained documents; for each of a plurality of the generated conditioning inputs, processing a network input generated from (i) the question data and (ii) the conditioning input using a neural network to generate a network output representing a candidate answer to the question; and generating, from the network outputs representing respective candidate answers, answer data representing a final answer to the question.

Abstract: A computer-implemented method, a computer program product, and a computer processing system are provided for online learning for a Dynamic Boltzmann Machine (DyBM) with hidden units. The method includes imposing, by a processor device, limited connections in the DyBM where (i) a current observation x[t] depends only on latest hidden units h[t-1/2] and all previous observations x[<t] and (ii) the latest hidden units h[t-1/2] depend on all the previous observations x[<t] while being independent of older hidden units h[t-1/2]. The method further includes computing, by the processor device, gradients of an objective function. The method also includes optimizing, by the processor device, the objective function in polynomial time using a stochastic Gradient Descent algorithm applied to the gradients.

Abstract: A method, system, and computer program product, including generating, using a linear probe, confidence scores through flattened intermediate representations and theoretically-justified weighting of samples during a training of the simple model using the confidence scores of the intermediate representations.

Abstract: A method, apparatus and computer program product for establishing a logical connection between an indirect utterance and a transaction is described. An indirect utterance is received from a user as an input to a conversational system. The indirect utterance is parsed to a first logical form. A first set of predicates and terms is mapped from the first logical form to a first subgraph in a knowledge graph. A second set of predicates and terms is mapped from a second logical form belonging to a transaction to a second subgraph of the knowledge graph. A best path in the knowledge graph between the first subgraph and the second subgraph is searched for while transforming the first logical form using the node and edge labels along the best path to generate an intermediate logical form. A system action is performed for a transaction if a graph structure of the intermediate logical form matches the graph structure of the logical form of the transaction above a threshold.

Abstract: A computer-implemented method and system having computer-executable instructions stored in a memory for processing user behavior features by neural networks to identify user segments. The method includes receiving user datasets from a database along with respective user identifiers, retention labels, static user features and interactive user features associated with an online product during a time period. A first neural network processes the interactive user features to generate a time distributed concatenation representation. A second neural network is configured to generate a vector by embedding the time distributed concatenation representation and the static user features through an embedding layer. The second neural network is configured to process the vector through a plurality of layers. A cluster model is used to determine user segments based on values extracted from nodes of a second to last layer of the second neural network.

Abstract: An artificial intelligence (AI) job recommender system and methods implement neural network machine learning by generating and utilizing actual and synthetic training data to identify, learn, and apply latent job-to-job transition information and trends to improve job recommendations. The AI job recommender system and method represent technological advances that, for example, identify data representations, identify multiple instances of latent information in actual data, develop synthetic training data, create a directed graph from latent, directional information, embed the directed graph into a vector space, and apply machine learning algorithms to technologically advance and transform a machine into a specialized machine that learns and improves job recommendations across the vector space.

Abstract: The disclosed technology relates identifying causes of an observed outcome. A system is configured to receive an indication of a user experience problem, wherein the user experience problem is associated with observed operations data including an observed outcome. The system generates, based on the observed operations data, a predicted outcome according to a model, determines that the observed outcome is within range of the predicted outcome, and identifies a set of candidate causes of the user experience problem when the observed outcome is within range of the predicted outcome.

Abstract: An exemplary embodiment provides an autoencoder which is explainable. An exemplary autoencoder may explain the degree to which each feature of the input attributed to the output of the system, which may be a compressed data representation. An exemplary embodiment may be used for classification, such as anomaly detection, as well as other scenarios where an autoencoder is input to another machine learning system or when an autoencoder is a component in an end-to-end deep learning architecture. An exemplary embodiment provides an explainable generative adversarial network that adds explainable generation, simulation and discrimination capabilities. The underlying architecture of an exemplary embodiment may be based on an explainable or interpretable neural network, allowing the underlying architecture to be a fully explainable white-box machine learning system.

Abstract: The present disclosure relates to systems, non-transitory computer-readable media, and methods that utilize recurrent neural networks to determine the existence of one or more open intents in a text input, and then extract the one or more open intents from the text input. In particular, in one or more embodiments, the disclosed systems utilize a trained intent existence neural network to determine the existence of an actionable intent within a text input. In response to verifying the existence of an actionable intent, the disclosed systems can apply a trained intent extraction neural network to extract the actionable intent from the text input. Furthermore, in one or more embodiments, the disclosed systems can generate a digital response based on the intent identified from the text input.

Abstract: A building system including one or more storage devices storing instructions thereon that, when executed by one or more processors, cause the one or more processors to receive an indication to execute an artificial intelligence (AI) agent of a digital twin, the digital twin including the AI agent and a graph data structure, the graph data structure including nodes representing entities of a building and edges between the nodes representing relationships between the entities of the building. The instructions cause the one or more processors to execute the AI agent based on data of the building to generate an inference that is a prediction of a future data value of a data point of the building for a future time and store at least one of the inference, or a link to the inference, in the graph data structure.

Abstract: Methods, systems, and devices for distributed hyperparameter tuning and load balancing are described. A device (e.g., an application server) may generate a first set of combinations of hyperparameter values associated with training a mathematical model. The mathematical model may include a machine learning model, an optimization model, or any combination. The device may identify a subset of combinations from the first set of combinations that are associated with a computational runtime that exceeds a first threshold and may distribute the subset of combinations across a set of machines. The device may then test each of the first set of combinations in a parallel processing operation to generate a first set of validation error values and may test a second set of combinations of hyperparameter values using an objective function that is based on the first set of validation error values.