Abstract: A machine learning system that uses a split net configuration to incorporate arbitrary constraints receives a set of input data and a set of functional constraints. The machine learning system jointly optimizes a deep learning model by using the set of input data and a wide learning model by using the set of constraints. The deep learning model includes an input layer, an output layer, and an intermediate layer between the input layer and the output layer. The wide learning model includes an input layer and an output layer but no intermediate layer. The machine learning system provides a machine learning model comprising the optimized deep learning model and the optimized wide learning model.

Abstract: Provided is a method and system with deep learning model generation. The method includes identifying a plurality of connections in a neural network that is pre-associated with a deep learning model, generating a plurality of pruned neural networks by pruning different sets of one or more of the plurality of connections to respectively generate each of the plurality of pruned neural networks, generating a plurality of intermediate deep learning models by generating a respective intermediate deep learning model corresponding to each of the plurality of pruned neural networks, and selecting one of the plurality of intermediate deep learning models, having a determined greatest accuracy among the plurality of intermediate deep learning models, to be an optimized deep learning model.

Abstract: A generative neural network control system controls a generative neural network by modifying the intermediate latent space in the generative neural network. The generative neural network includes multiple layers each generating a set of activation values. An initial layer (and optionally additional layers) receives an input latent vector, and a final layer outputs an image generated based on the input latent vector. The data that is input to each layer (other than the initial layer) is referred to as data in an intermediate latent space. The data in the intermediate latent space includes activation values (e.g., generated by the previous layer or modified using various techniques) and optionally a latent vector. The generative neural network control system modifies the intermediate latent space to achieve various different effects when generating a new image.

Abstract: Techniques that facilitate machine learning using multi-dimensional time series data are provided. In one example, a system includes a snapshot component and a machine learning component. The snapshot component generates a first sequence of multi-dimensional time series data and a second sequence of multi-dimensional time series data from multi-dimensional time series data associated with at least two different data types generated by a data system over a consecutive period of time. The machine learning component that analyzes the first sequence of multi-dimensional time series data and the second sequence of multi-dimensional time series data using a convolutional neural network system to predict an event associated with the multi-dimensional time series data.

Abstract: A control device executes the following processes: a first process of detecting that autonomous traveling control is not able to be continued and executing stop control of a vehicle; a second process of determining whether a person is present in the vehicle; and a third process of unlocking a door lock of the vehicle when at least one person is present in the vehicle after the vehicle is stopped by the first process.

Abstract: A method for providing an intelligent electronic communication to a consumer device using a recommendation engine.

Abstract: An example operation includes one or more of determining a sensor on a transport is not functioning properly, determining a severity of the malfunction, responsive to the severity exceeding a threshold, lowering an autonomous level of the transport, and responsive to the severity continuing to exceed the threshold, limiting an operation of the transport based on an intended output of the malfunctioning sensor.

Abstract: In various examples, a neural network may be trained for use in vehicle re-identification tasks—e.g., matching appearances and classifications of vehicles across frames—in a camera network. The neural network may be trained to learn an embedding space such that embeddings corresponding to vehicles of the same identify are projected closer to one another within the embedding space, as compared to vehicles representing different identities. To accurately and efficiently learn the embedding space, the neural network may be trained using a contrastive loss function or a triplet loss function. In addition, to further improve accuracy and efficiency, a sampling technique—referred to herein as batch sample—may be used to identify embeddings, during training, that are most meaningful for updating parameters of the neural network.

Abstract: With rapidly evolving technologies and emerging tools, sports-related videos generated online are rapidly increasing. To automate the sports video editing/highlight generation process, a key task is to precisely recognize and locate events-of-interest in videos. Embodiments herein comprise a two-stage paradigm to detect categories of events and when these events happen in videos. In one or more embodiments, multiple action recognition models extract high-level semantic features, and a transformer-based temporal detection module locates target events. These novel approaches achieved state-of-the-art performance in both action spotting and replay grounding. While presented in the context of sports, it shall be noted that the systems and methods herein may be used for videos comprising other content and events.

Abstract: A battery information processing system having: an acquisition unit configured to acquire battery information including use histories of batteries, a classification unit configured to classify the batteries for each property on the basis of properties of the batteries and store the battery information in a plurality of databases classified for each property, a selection unit configured to select a database including high priority battery information from the plurality of classified databases, a determination unit configured to determine the selected database as a database for selecting a combination of a plurality of batteries in a case in which a property of an assembled battery calculated by combining a plurality of batteries satisfies a criterion, and a presentation unit configured to present combination information of the plurality of batteries selected in a descending order from the highest coincidence.

Abstract: In one aspect, at least one device may include at least one processor and storage accessible to the at least one processor. The storage may include instructions executable by the at least one processor to access an image and to execute object recognition using the image to identify at least a first object from the image. The instructions may also be executable to access a database indicating whether use permissions are grantable for various objects and to determine, based on accessing the database, that use permissions are grantable for the first object. The instructions may then be executable to control a display to present a graphical user interface (GUI) on the display, where the GUI may indicate the use permissions for the first object and include a selector that is selectable to initiate secure communication to authenticate an entity as being granted the use permissions for the first object.

Abstract: Embodiments relate to a neural processor circuit that may include a fetch circuit that fetches coefficient data of a machine learning model from a memory source. The neural processor circuit may also include one or more neural engine circuits that are coupled to the fetch circuit. A neural engine circuit may include a buffer circuit that stores the coefficient data. The neural engine circuit may also include a coefficient organizing circuit that generates at least a first mapping and a second mapping of the stored coefficient data according to one or more control signals. The neural engine may also include a computation circuit that receives and processes at least a portion of input data with the coefficient data as mapped according to the first mapping or process at least the portion of the input data with the coefficient data as mapped according to the second mapping.

Abstract: Generating expanded responses that guide continuance of a human-to computer dialog that is facilitated by a client device and that is between at least one user and an automated assistant. The expanded responses are generated by the automated assistant in response to user interface input provided by the user via the client device, and are caused to be rendered to the user via the client device, as a response, by the automated assistant, to the user interface input of the user. An expanded response is generated based on at least one entity of interest determined based on the user interface input, and is generated to incorporate content related to one or more additional entities that are related to the entity of interest, but that are not explicitly referenced by the user interface input.

Abstract: An electronic device and a method for generating a reference configuration of a computing device are provided. The method includes: obtaining a first neural network model, wherein the first neural network model includes a plurality of cluster centers, wherein the plurality of cluster centers correspond to a plurality of features; obtaining a first configuration requirement; determining that the first configuration requirement corresponding to a first cluster center among the plurality of cluster centers; generating the reference configuration according to a plurality of first feature values of the first cluster center, wherein the plurality of first feature values correspond to the plurality of features, respectively; and outputting the reference configuration.

Abstract: Techniques are provided for improving image data quality, such as in functional imaging follow-up studies, using reconstruction, post-processing, and/or deep-learning enhancement approaches in a way that automatically improves analysis fidelity, such as lesion tracking fidelity. The disclosed approaches may be useful in improving the performance of automatic analysis methods as well as in facilitating reviews performed by clinician.

Abstract: An arithmetic operation circuit including: a variable resistance element that includes three terminals that are a first terminal, a second terminal, and a third terminal and is configured to be able to change a resistance value; a first electrode connected to the first terminal; a second electrode; a third electrode; a first switching element connected between the second electrode and the second terminal; a second switching element connected between the third electrode and the third terminal; and a capacitor connected between a transmission line connecting the second terminal and the first switching element and the ground.

Abstract: Systems and methods for identifying molecules that are biologically active against a disease, where the method can comprise culturing a first mammalian cell population under organoid formation conditions in the presence of a test molecule to obtain a first organoid, wherein the first mammalian cell population, when cultured under the organoid formation conditions in the absence of the test molecule, results in an organoid with a disease phenotype; imaging the first organoid following exposure to the test molecule; analyzing one or more images of the first organoid using a neural network that has been trained to assign a probability score of disease or non-disease ranging between 0% and 100%; assigning the first organoid a probability score ranging between 0% and 100%; wherein the test molecule is biologically active against the disease if the probability score of the first organoid is greater than a cutoff probability score of non-disease or lower than a cutoff probability score of disease.

Abstract: A data structuring system that provides a user interface to enable data wrangling and modeling, and methods for making and using the same.

Abstract: The disclosure herein describes training a global model based on a plurality of data sets. The global model is applied to each data set of the plurality of data sets and a plurality of gradients is generated based on that application. At least one gradient quality metric is determined for each gradient of the plurality of gradients. Based on the determined gradient quality metrics of the plurality of gradients, a plurality of weight factors is calculated. The plurality of gradients is transformed into a plurality of weighted gradients based on the calculated plurality of weight factors and a global gradient is generated based on the plurality of weighted gradients. The global model is updated based on the global gradient, wherein the updated global model, when applied to a data set, performs a task based on the data set and provides model output based on performing the task.

Abstract: Some embodiments provide a method for training multiple parameters of a machine-trained (MT) network subject to a sparsity constraint that requires a threshold portion of the parameters to be equal to zero. A first set of the parameters subject to the sparsity constraint are grouped into groups of parameters. For each parameter of a second set of the parameters subject to the sparsity constraint, the method determines an accuracy penalty associated with the parameter being set to zero. For each group of parameters in the first set of parameters, the method determines a minimum accuracy penalty for each possible number of parameters in the group being set to zero. The method uses the determined accuracy penalties to set to the value zero at least the threshold portion of the plurality of parameters.

Abstract: To take advantage of the architecture of a systolic array tailored to perform sparse matrix multiplications, a weight matrix can be converted into a set of constrained fine-grained sparse weight matrices. The conversion process may include receiving a request to perform a matrix multiplication operation with a weight matrix, and determining that the weight matrix satisfies a sparsity condition to convert the weight matrix into a set of constrained fine-grained sparse weight matrices. The weight matrix can then be converted into a set of constrained fine-grained sparse weight matrices. Computer instructions can then be generated for an integrated circuit device to perform the requested matrix multiplication operation as a set of sparse matrix multiplication operations using the set of constrained fine-grained sparse weight matrices.

Abstract: Methods and computer program products for training a neural network perform multiple forms of data augmentation on sample waveforms of a training dataset that includes both normal and abnormal samples to generate normal data augmentation samples and abnormal data augmentation samples. The normal data augmentation samples are labeled according to a type of data augmentation that was performed on each respective normal data augmentation sample. The abnormal data augmentation samples are labeled according to a type of data augmentation other than that which was performed on each respective abnormal data augmentation sample. A neural network model is trained to identify a form of data augmentation that has been performed on a waveform using the normal data augmentation samples and the abnormal data augmentation samples.

Abstract: Modelling for efficient resource allocation and/or distribution in secure computation and communication for differential privacy protocols and/or algorithms is provided. A method for allocating a resource of a differentially private system in secure computation and communication includes aggregating attributes from a usage log of the differentially private system for a predetermined period of time, generating a moving aggregation based on the aggregated attributes, training a machine learning model based on the aggregated attributes and the moving aggregation, predicting a distribution of the resource using the trained machine learning model, and allocating the resource based on the predicted distribution. The resource includes a differential privacy parameter.

Abstract: An arithmetic device includes a multiplying-accumulating (MAC) operator and an activation function (AF) circuit. The MAC operator performs a MAC arithmetic operation for weight data and vector data to generate an arithmetic result signal. The AF circuit extracts a first bit group and a second bit group from the arithmetic result signal. In addition, the AF circuit generates an input distribution signal based on the first bit group and the second bit group. Moreover, the AF circuit selects and outputs an output distribution signal that corresponds to the input distribution signal based on an activation function.

Abstract: Aspects of the present invention disclose a method, computer program product, and system for query execution in a multi-tenant cloud service. The method includes one or more processors determining category classes for service queries. The method further includes sending for execution, a selected number of service queries from one of the determined category classes to a shadow query engine. Respective service queries of the categorically classified service queries comprise a different set of configuration parameter values for the shadow query engine. The method further includes recording metadata for the selected number of service queries of the one category class executed on said shadow query engine. The method further includes determining correlations between the recorded metadata.

Abstract: Static parameters of a software container are identified that relate to metadata of the software container itself. The software container is assigned to a selected runtime environment based on the static parameters using a first machine learning model. Runtime parameters for the software container are identified by analyzing the software container at runtime. The runtime parameters relate to operations that the software container requires during runtime. Using a second machine learning model, it is determined whether the selected runtime environment matches the runtime parameters. Where the runtime environment matches, the software container continues to run in this environment. Where the runtime environment does not match, the software container is run in a different runtime environment that matches both the static and runtime parameters.

Abstract: Disclosed are a data loading method and apparatus for a convolution operation, a computer device, a storage medium and a computer program product. The method includes: splitting a feature image to be loaded into the cache unit into a plurality of sub-feature images; determining a target cache line corresponding to each of the sub-feature images in each of the cache lines according to a positional relationship of each of the sub-feature images in the feature image; wherein target cache lines corresponding to at least two sub-feature images with the same positional relationship are located in the same cache set, and target cache lines corresponding to at least two sub-feature images with an adjacent positional relationship are located in different cache sets; loading a data content of each of the sub-feature images into the target cache line corresponding to each of the sub-feature images.

Abstract: An analysis engine receives data characterizing a prompt for ingestion by a generative artificial intelligence (GenAI) model. The analysis engine, using a determines using, for example, a classifier or blocklist, that the prompt comprises or is indicative of malicious content or otherwise elicits undesired model behavior. Similarly, outputs of the GenAI model can be analyzed to determine whether they comprise malicious content or cause the model to behave in an undesired manner. The output is inputted into a GenAI model along with obfuscation instructions to generate an output which is returned to the requesting user. Related apparatus, systems, techniques and articles are also described.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for distributing machine learning workloads, e.g., computations for training a neural network or computing an inference using a neural network, across multiple hardware accelerators.

Abstract: Systems and methods for the control and operation of a grid-connected converter with an energy storage system are provided. The system can include a small microgrid comprising an AC grid that is feeding a DC load through a converter. The converter is connected to the AC grid through an R-L filter. The classical linear controllers have limitations due to their slow transient performance and low robustness against parameter variations and load disturbances. In certain embodiments, the transient and steady state responses of the provided artificial intelligence based Robust Artificial Neural Network Tracking Control (RANNTC) of three-phase grid-connected power converters has been shown to be more enhanced in terms of overshoot (24% lower), settling time (85% reduced), and total harmonic distortion (THD) (55% lower).

Abstract: A device may receive unprocessed images to be labeled, and may utilize a first neural network model to identify objects of interest in the unprocessed images and bounding boxes for the objects of interest. The device may annotate the objects of interest to generate annotated objects of interest, and may utilize a second neural network model to group the annotated objects of interest into clusters. The device may utilize a third neural network model to determine labels for the clusters, and may request manually-generated labels for clusters for which labels are not determined. The device may receive the manually-generated labels, and may label the unprocessed images with the labels and the manually-generated labels to generate labeled images. The device may generate a training dataset based on the labeled images, and may train a computer vision model with the training dataset to generate a trained computer vision model.

Abstract: A method of object re-identification in images of objects comprises providing a plurality of neural networks for object re-identification, wherein each of the plurality of neural networks is trained on image data with different sets of anatomical features, each set being represented by a reference vector; receiving a plurality of images of objects and an input vector representing anatomical features that are depicted in all of the plurality of images; comparing the input vector with the reference vectors for determining, according to a predefined condition, the most similar reference vector; and inputting image data of the plurality of objects to the neural network represented by the most similar reference vector for determining whether the plurality of objects have the same identity.

Abstract: The technical solution involves a board card including a storage component, an interface apparatus, a control component, and an artificial intelligence chip. The artificial intelligence chip is connected to the storage component, the control component, and the interface apparatus, respectively; the storage component is used to store data; the interface apparatus is used to implement data transfer between the artificial intelligence chip and an external device; and the control component is used to monitor a state of the artificial intelligence chip. The board card is used to perform an artificial intelligence operation.

Abstract: A method includes identifying a plurality of data items responsive to a first query. A presentation value associated with each of the plurality of data items is determined, the presentation value of a respective data item being a value associated with the respective data item by a first user in exchange for the presentation of the data item by a publication system. The plurality of data items are ranked for presentation to a second user, the ranking being performed using the respective presentation values associated with the plurality of data items.

Abstract: Systems, methods, and apparatuses for providing dynamic, prioritized spectrum utilization management. The system includes at least one monitoring sensor, at least one data analysis engine, at least one application, a semantic engine, a programmable rules and policy editor, a tip and cue server, and/or a control panel. The tip and cue server is operable utilize the environmental awareness from the data processed by the at least one data analysis engine in combination with additional information to create actionable data.

Abstract: The subject disclosure relates to solutions for reducing or eliminating motion sickness experienced by a vehicle occupant/passenger. In some aspects, a process of the disclosed technology includes steps for collecting motion data associated with a vehicle using one or more environmental sensors, tracking eye movements of a user within a cabin of the vehicle, processing the motion data and the eye movements to identify a motion event, and generating a motion compensation signal based on the motion event. Systems and machine-readable media are also provided.

Abstract: Systems and methods for monitoring a fleet of self-driving vehicles are disclosed. The system comprises one or more self-driving vehicles having at least one sensor for collecting current state information, a fleet-management system, and computer-readable media for storing reference data. The method comprises autonomously navigating a self-driving vehicle in an environment, collecting current state information using the vehicle's sensor, comparing the current state information with the reference data, identifying outlier data in the current state information, and generating an alert based on the outlier data. A notification based on the alert may be sent to one or more monitoring devices according to the type and severity of the outlier.

Abstract: An information processing apparatus includes a processor including hardware. The processor extracts neighboring nodes in two or more different extraction ranges for each node constituting input data of a graph structure. The processor calculates an anomaly score representing a degree of anomaly of the node for each extraction range based on a representation of a combination of the node and the neighboring nodes. The processor records each calculated anomaly score in a storage. The processor selects a maximum anomaly score among the anomaly scores recorded in the storage. The processor determines an anomaly node in the input data of the graph structure based on the selected maximum anomaly score. The processor outputs information of the anomaly node.

Abstract: A method for automated detection of cervical pre-cancer includes: providing at least one cervigram; pre-processing the at least one cervigram; extracting features from the at least one pre-processed cervigram; and classifying the at least one cervigram as negative or positive for cervical pre-cancer based on the extracted features.

Abstract: A control section is included, the control section including an accumulation function of, when recognizing a specific user on a basis of sensor data acquired via an agent device, generating episode data in the accumulation section on a basis of a keyword extracted from the sensor data, generating a question for drawing out information concerning the episode data, and accumulating a reply from the specific user to the question in the episode data, and a responding function of, when recognizing the specific user on the basis of the sensor data acquired via the agent device, retrieving the episode data through the accumulation section on the basis of the keyword extracted from the sensor data, and generating response data concerning the retrieved episode data for the agent device to respond to the specific user.

Abstract: The present teaching is directed to clustering with denoising capability and its use in network capacity planning. Data samples with attributes of network elements and respective key performance indicators are first clustered to obtain initial clusters. Each initial cluster is hierarchically clustered to generate subclusters, each of which is detected as a pure or an impure subcluster based on some criterion. Each impure subcluster is denoised based on a situation detected, with some samples merged with a corresponding pure subcluster, some bootstrapped using additional data samples with consistent properties, and some removed if additional data sample with consistent properties is not available. The denoising is iteratively performed until a denoising criterion is satisfied to obtain denoised clusters corresponding to clusters of network elements. Actions may be performed on the network elements according to their corresponding denoised clusters.

Abstract: Provided are a method of performing a convolution operation between a kernel and an input feature map based on reuse of the input feature map, and a neural network apparatus using the method. The neural network apparatus generates output values of an operation between each of weights of a kernel and an input feature map, and generates an output feature map by accumulating the output values at positions in the output feature map that are set based on positions of the weights in the kernel.

Abstract: Aspects of the present disclosure involve implementations that may be used to protect neural network models against adversarial attacks by obfuscating neural network operations and architecture. Obfuscation techniques include obfuscating weights and biases of neural network nodes, obfuscating activation functions used by neural networks, as well as obfuscating neural network architecture by introducing dummy operations, dummy nodes, and dummy layers into the neural networks.

Abstract: A computer-implemented method for assessing a person re-identification risk in an application domain is provided. In the application domain, for each of a plurality of persons a corresponding personal record is stored in a database. Each record comprises a set of attributes. Each attribute comprises a corresponding attribute name and a corresponding attribute value.

Abstract: This learning device provides a learned model to an adjuster including the learned model learned to output a predetermined compensation amount to a controller based on parameters of an object to be processed, in a system including the controller outputting a command value obtained by compensating a target value based on a compensation amount; and a control object performing a predetermined process on the object and outputting a control variable as a response to the command value. The learning device includes: a learning part generating candidate compensation amounts based on operation data including a target value, command value and control variable, learning with the generated candidate compensation amounts and the parameters of the object as teacher data, and generating or updating the learned model; and a setting part providing, to the adjuster, the generated or updated learned model.

Abstract: For reconstruction in medical imaging using a scan protocol with repetition, a machine learning model is trained for reconstruction of an image for each repetition. Rather than using a loss for that repetition in training, the loss based on an aggregation of images reconstructed from multiple repetitions is used to train the machine learning model. This loss for reconstruction of one repetition based on aggregation of reconstructions for multiple repetitions is based on deep set-based deep learning. The resulting machine-learned model may better reconstruct an image from a given repetition and/or a combined image from multiple repetitions than a model learned from a loss per repetition.

Abstract: To perform a beam search operation on an input tensor using a data processor with native hardware support, the data processor can be programmed with a set of instructions. The set of instructions can include a first machine instruction that operates on the input tensor to obtain N largest values in the input tensor, a second machine instruction that operates on the input tensor to obtain indices corresponding to the N largest values in the input tensor, and a third machine instruction that operates on the input tensor to replace the N largest values in the input tensor with a minimum value.

Abstract: A neural processing unit (NPU) for testing a component during runtime is provided. The NPU may include a plurality of functional components including a first functional component and a second functional component. At least one of the plurality of functional components may be driven for calculation of an artificial neural network. Another one of the plurality of functional components may be selected as a component under test (CUT). A scan test may be performed on the at least one functional component selected as the CUT. A tester for detecting a defect of an NPU is also provided. The tester may include a component tester configured to communicate with at least one functional component of the NPU, select the at least one functional component as a CUT, and perform a scan test for the selected CUT.

Abstract: Variational Autoencoders (VAEs) have been shown to be effective in modeling complex data distributions. Conventional VAEs operate with fully-observed data during training. However, learning a VAE model from partially-observed data is still a problem. A modified VAE framework is proposed that can learn from partially-observed data conditioned on the fully-observed mask. A model described in various embodiments is capable of learning a proper proposal distribution based on the missing data. The framework is evaluated for both high-dimensional multimodal data and low dimensional tabular data.

Abstract: A method of training one or more neural networks, the one or more neural networks being for use in lossy image or video encoding, transmission and decoding, the method comprising the steps of: receiving an input image at a first computer system; encoding the input image using a first neural network to produce a latent representation; decoding the latent representation using a second neural network to produce an output image, wherein the output image is an approximation of the input image; evaluating a function based on a difference between the output image and the input image; updating the parameters of the first neural network and the second neural network based on the evaluated function; and repeating the above steps using a first set of input images to produce a first trained neural network and a second trained neural network; wherein the difference between the output image and the input image is determined based on the output of a neural network acting as a discriminator; the parameters of the neural netw