Abstract: A receiver receives, from a transmitter, a reference neural network. The receiver trains the reference neural network to obtain updated neural network parameters for the reference neural network. The receiver reports to the transmitter in response to a trigger, a difference between the updated neural network parameters and previous neural network parameters for the reference neural network. The trigger may be based on a loss function, a magnitude of the difference between the updated neural network parameters and the previous neural network parameters, and/or a difference between performance of the reference neural network with the updated neural network parameters and performance of the reference neural network with the previous neural network parameters.

Abstract: A computer-implemented method for correcting data to match user identity may include (i) receiving user input specifying an aspect of physical presentation of the user that does not match an authentic identity of the user, where the authentic identity of the user includes a realistic version of the user that reflects an internal self-image of the user, (ii) capturing, via a sensor, data of the user that includes the aspect of the physical presentation of the user, (iii) correcting the captured data of the user to portray a corrected version of the aspect that matches the authentic identity of the user, and (iv) storing the corrected data of the user that matches the authentic identity of the user instead of uncorrected data of the user that includes the aspect that does not match the authentic identity of the user. Various other methods, systems, and computer-readable media are also disclosed.

Abstract: Disclosed are an apparatus and a method for neural network learning using a synapse based on multiple elements. A neural network learning apparatus using a synapse based on multiple elements according to an exemplary embodiment of the present disclosure includes a first synaptic unit including a plurality of first resistive elements to update a weight of a neural network based on a first precision and a second synaptic unit including a plurality of second resistive elements to update the weight of the neural network with a precision higher than the first precision.

Abstract: A method of fault criticality assessment using neural twins includes converting a netlist into a neural twin by replacing each circuit element of the netlist with a neural-network-readable cell equivalent; and replacing each wire with a neural connection. Bias value adders are inserted at locations in the neural twin; and these bias value adders are used to apply a bias that represents a perturbation in the signal propagated by that connection. For each perturbed bias at a corresponding site selected to be perturbed, a loss value is calculated for the neural twin; and the site is classified, using a neural-twin-trained classifier, as critical or benign based on that loss value.

Abstract: A method of multi-rate neural image compression is performed by at least one processor and includes selecting encoding masks, based on a first hyperparameter, and performing a convolution of a first plurality of weights of a first neural network and the selected encoding masks to obtain first masked weights. The method further includes encoding an input image to obtain an encoded representation, using the first masked weights, and encoding the obtained encoded representation to obtain a compressed representation.

Abstract: The present disclosure provides an adaptive artificial-intelligence framework for modeling, synthesizing, and evaluating relationships among multiple discrete entities. The system aggregates inputs into composite relation entities (CREs). A CRE is a machine-enforced identity object with inseparable structural metadata, and is not a simple embedding, vector, or record.

Abstract: Convolutional operators for triangle meshes are determined to construct one or more neural networks. In at least one embodiment, convolutional operators, pooling operators, and unpooling operators are determined to construct the one or more neural networks, in which the same learned weights from the one or more neural networks can further be used for triangle meshes with different topologies.

Abstract: Disclosed herein are related to a device for performing neuromorphic computing. In one aspect, a device includes a back end of line layer including a three-dimensional memory array. The three-dimensional memory array may include a plurality of memory cells to store a plurality sets of weight values of a neural network model. In one aspect, the device includes a front end of line layer including a controller. The controller may apply one or more input voltages corresponding to an input to the neural network model to the three-dimensional memory array, and receive one or more output voltages from the three-dimensional memory array to perform computations of the neural network model.

Abstract: Various embodiments analyze the application of satellite imaging and deep learning in predicting ignition and spread of major wildfires. The training data comes from NASA satellite products and historical records of wildfires in the United States. A state-of-the-art technique in neural network image classification may be utilized and yield impressive results for wildfire ignition prediction. In one embodiment, the model may achieve an accuracy rate of 93.5%, a precision rate of 93.2%, a recall rate of 88.9%, and an F-1 score of 90.8%. Direct applications of the model may include wildfire monitoring and wildfire prevention.

Abstract: The acceptance means 81 accepts input of a series of observed events. The deletion means 82 deletes, in a decision graph for determining a boolean value of the observed event based on a knowledge base in which the observed event is a node and transition relationship of each of the observed events is an edge, the edge indicating a transition to a negation side of the observed event from among the edges connected to the node corresponding to the accepted observed event. The determination means 83 determines whether or not a path from a root node to a true-valued leaf node exists in the decision graph after the deletion of the edge. The decision means 84 decides that the series of observed events is inconsistent with respect to the knowledge base when being determined that there is no path to the leaf node indicating the true value.

Abstract: A system and method for predicting next pitch are disclosed herein. A computing system retrieves pitch-by-pitch information for a plurality of events and game context information associated with each pitch in the pitch-by-pitch information. The computing system converts the pitch-by-pitch information and the game context information into a plurality of graph-based representation. A graph neural network learns to generate a pitch type prediction for each pitch based on the plurality of graph-based representations. The computing system generates a trained graph neural network based on the learning. The computing system receives a current graph-based representation of current pitch-by-pitch information for a current pitcher and current game context information. The computing system predicts, via the trained graph neural network, a pitch type for the next pitch to be delivered from the current pitcher.

Abstract: A computer-implemented method is provided. The method may comprise: obtaining at least one document to be classified; classifying, using a machine learning model including an artificial neural network (ANN) and an attention mechanism, the at least one document into at least two classes; determining, for each of the at least one document, a confidence value of the classifying, based on one or more outputs of one or more nodes comprised in the ANN; assigning, to each of the at least one document, based at least in part on the confidence value, one of at least two categories that are associated with different degrees of credibility of the classifying; and providing for display one or more of the at least one document with: the assigned category and attention information that indicates significance of one or more parts of each document provided for display in the classifying of said document.

Abstract: Systems and methods for generating a hypernetwork configured to be trained for a plurality of tasks; receiving a task preference vector identifying a hierarchical priority for the plurality of tasks, and a resource constraint as a tuple; finding tree sub-structures and the corresponding modulation of features for every tuple within an N-stream anchor network; optimizing a branching regularized loss function to train an edge hypernet; and training a weight hypernet, keeping the anchor net and the edge hypernet fixed.

Abstract: Systems and methods for retrieving similar multivariate time series segments are provided. The systems and methods include extracting a long feature vector and a short feature vector from a time series segment, converting the long feature vector into a long binary code, and converting the short feature vector into a short binary code. The systems and methods further include obtaining a subset of long binary codes from a binary dictionary storing dictionary long codes based on the short binary codes, and calculating similarity measure for each pair of the long feature vector with each dictionary long code. The systems and methods further include identifying a predetermined number of dictionary long codes having the similarity measures indicting a closest relationship between the long binary codes and dictionary long codes, and retrieving a predetermined number of time series segments associated with the predetermined number of dictionary long codes.

Abstract: One or more computer processors create a fully convolution network (FCN) comprising a plurality of 1×1 convolutions. The one or more computer processors append linear mapping layer (LM) to created FCN. The one or more computer processors capture a plurality of features utilizing multi-scale dilated convolutional kernels from the linear mapped FCN (LM-FCN). The one or more computer processors apply an average pool layer to the captured plurality of features along a temporal axis of a dilated convolutional kernel within the LM-FCN. The one or more computer processors predict a classification for subsequent time-series data utilizing the pooled plurality of features.

Abstract: A computer implemented method includes identifying, by one or more processors, a decision tree corresponding to an artificial intelligence model, detecting, by one or more processors, new data associated with an update to the identified decision tree, identifying, by one or more processors, counterfactual data corresponding to the new data, identifying, by one or more processors, one or more expected outcomes corresponding to the counterfactual data and the new data, and generating, by one or more processors, an updated decision tree based on the identified new data and the identified counterfactual data. A computer program product and computer system corresponding to the method are also disclosed.

Abstract: One example method includes deploying, in a production environment, a machine learning model that was trained using metadata created by an intermediate classical computing layer, and the metadata comprises information about one or more aspects of a quantum circuit, generating, with the machine learning model, a prediction as to how one or more computing infrastructures may be expected to perform when executing the quantum circuit, based on the prediction, making an orchestration decision concerning the quantum circuit, and orchestrating the quantum circuit to one of the computing infrastructures.

Abstract: This disclosure provides an accurate and personalized recommendation method based on a knowledge graph, which includes following steps: acquiring relevant knowledge of objects from a knowledge base according to historical behaviors of a user, and constructing a knowledge graph; initializing a vector representation of each node and its connection, and determining a receptive field of the node; generating training samples according to the historical behaviors of the user, and initializing a vector representation of all users and objects; acquiring a receptive field of an entity in the knowledge graph corresponding to the object in the training sample, then inputting the receptive field and the training sample to a graph neural network model to obtain predicted values of a possibility of an interaction between the user and the object.

Abstract: Systems and techniques that facilitate quantum-enhanced features for classical machine learning are provided. In various embodiments, a system can comprise a receiver component that can access a classical dataset. In various aspects, the system can further comprise a feature component that can generate one or more machine learning input features based on a quantum transformation of the classical data set. In various instances, the system can further comprise an execution component that can execute a classical machine learning model on the one or more machine learning input features.

Abstract: A computer implemented method obtains neural network-based model base model weight matrices for each of multiple neural network layers. First low-rank factorization matrices are added to corresponding base model weight matrices to form a first domain model. The low-rank factorization matrices are treated as trainable parameters. The first domain model is trained with first domain specific training data without modifying base model weight matrices.