Abstract: Embodiments relate to a neural processor circuit including one or more planar engine circuits that perform non-convolution operations in parallel with convolution operations performed by one or more neural engine circuits. The neural engine circuits perform the convolution operations on neural input data corresponding to one or more neural engine tasks to generate neural output data. The planar engine circuits perform non-convolution operations on planar input data corresponding to one or more planar engine tasks to generate planar output data. A data processor circuit that includes multiple buffer circuits performs task skew management between the one or more neural engine tasks and the one or more planar engine tasks. The data processor circuit stops addition of an incoming task to queues in response to one or more of the queues stored in the buffer circuits reaching a threshold.

Abstract: A method and system are provided. The method includes receiving, at a generator, a random input, producing, at the generator, a synthetic output of the received random input, receiving, at a teacher network, the synthetic output, receiving, at a student network, the synthetic output, minimizing a maximum of a distance between an output of the teacher network and an output of the student network, and constraining the generator.

Abstract: A training method includes obtaining a first data set and a second data set, each of the first data set and the second data set including event instances, the event instances include text and events corresponding to the text. The training method also includes training an adversarial network using the first data set and the second data set, the adversarial network includes processing circuitry configured as a generator and a discriminator. The discriminator is configured to output first reliable probabilities of the event instances in the first data set, and second reliable probabilities of the event instances inputted by the generator. A loss function of the adversarial network is used to adjust a parameter of the adversarial network, to maximize the first reliable probabilities and minimize the second reliable probabilities. The method further includes obtaining, by the trained adversarial network, a reliable event instance in the second data set.

Abstract: Systems and methods are provided for performing customizable machine prediction using an extensible software tool. A specification including features of a trained machine learning model can be received and an interface for the trained machine learning model can be generated. The trained machine learning model can be loaded using the interface, the loaded machine learning model including a binary file configured to receive data as input and generate prediction data as output. Predictions can be generated using observed data that is stored according to a multidimensional data model, wherein a portion of the observed data is input to the loaded machine learning model to generate first data predictions, and a portion of the observed data is used by a generic forecast model to generate second data predictions. The first and second data predictions can be displayed in a user interface configured to display intersections of the multidimensional data model.

Abstract: An AI based defect inspecting device and method is disclosed. The present embodiment, in determining the good or defective product using the deep learning-based classification model based on an image of the product, provides a defect inspecting device and method for providing a basis for determining a good/defective product provided by a deep learning-based classification model using explainable AI (XAI) generating a category set for the basis, and continuously updating parameters of the deep learning-based classification model using the category set.

Abstract: This disclosure describes methods, non-transitory computer readable storage media, and systems that utilize a plurality of neural networks to determine structural and semantic information via different views of a word sequence and then utilize this information to extract a relationship between word sequence entities. For example, the disclosed systems generate a plurality of sets of encoded word representation vectors utilizing the plurality of neural networks. The disclosed system then extracts the relationship from an overall word representation vector generated based on the sets of encoded word representation vectors. Furthermore, the disclosed system enforces structural and semantic consistency between views via a plurality of constrains involving a control mechanism for the semantic view and a plurality of losses.

Abstract: Techniques are disclosed for using a machine learning model to identify and present a ranked array of interface elements representing entities. The location of individual interface elements within the ranked array of interface elements is based on a level of match between entity attributes and a set of requirements established by a user. The machine learning model may be further trained by receiving a user input that changes a location of a particular user interface element within a graphical user interface displaying the ranked array. Upon receiving the user input, the trained machine learning model may update training data to include an updated match score for the particular user interface element that reflects the new location.

Abstract: A method for executing a multi-task deep learning model for learning trends in multivariate time series is presented. The method includes collecting multi-variate time series data from a plurality of sensors, jointly learning both local and global contextual features for predicting a trend of the multivariate time series by employing a tensorized long short-term memory (LSTM) with adaptive shared memory (TLASM) to learn historical dependency of historical trends, and employing a multi-task one-dimensional convolutional neural network (1dCNN) to extract salient features from local raw time series data to model a short-term dependency between local time series data and subsequent trends.

Abstract: An analysis engine receives data characterizing a multimodal prompt for ingestion by a generative artificial intelligence (GenAI) model. The multimodal prompt is processed and fed into a plurality of layers from which an intermediate result of the GenAI model or a proxy of the GenAI model is obtained. The analysis engine, using a classifier and the intermediate result, determines whether the prompt elicits undesired behavior by the GenAI model. Data characterizing the determination is provided to a consuming application or process. Related apparatus, systems, techniques and articles are also described.

Abstract: Systems and methods for constructing layered prompts to operate as input into a pre-trained large language model (LLM). The method involves obtaining a set of application domains in which the LLM will be used. Using these application domains, a set of guidelines is determined, defining operation boundaries for the LLM. A set of layers is determined, each associated with the guidelines and including variables representing attributes identified within those guidelines. Using these layers, a first layered prompt is constructed to test the initial operation boundaries of the guidelines and is supplied to the LLM to generate a set of responses. Based on the responses, a second layered prompt is dynamically constructed to test additional operation boundaries, ensuring iterative refinement and contextual relevance.

Abstract: Embodiments described herein provide systems and methods for noise-robust contrastive learning. In view of the need for a noise-robust learning system, embodiments described herein provides a contrastive learning mechanism that combats noise by learning robust representations of the noisy data samples. Specifically, the training images are projected into a low-dimensional subspace, and the geometric structure of the subspace is regularized with: (1) a consistency contrastive loss that enforces images with perturbations to have similar embeddings; and (2) a prototypical contrastive loss augmented with a predetermined learning principle, which encourages the embedding for a linearly-interpolated input to have the same linear relationship with respect to the class prototypes. The low-dimensional embeddings are also trained to reconstruct the high-dimensional features, which preserves the learned information and regularizes the classifier.

Abstract: Techniques are provided for using machine learning techniques to learn embeddings for content items. In one technique, training data is used to learn embeddings for each attribute value of multiple attribute values of multiple content items, embeddings for each attribute value of multiple attribute values of multiple entities, and weights for a set of contextual features. In response to receiving a content request, a content item that is associated with one or more targeting criteria that are satisfied based on the content request is identified. A first set of embeddings for the content item are identified, a requesting entity that initiated the content request is identified along with a second set of embeddings for the requesting entity, and a set of feature values for the set of contextual features is identified. The content item is selected based on the sets of embeddings, the set of feature values, and the weights.

Abstract: Computing systems, computer-readable media, and methods for providing multiple computer-generated seismic data interpretation options, of which the method includes receiving a training input, sorting the training input into a first group and a second group, subgrouping the second group into a plurality of subgroups, generating a plurality of trained models based on the plurality of subgroups and the first group, receiving a prediction input having a set of data to be interpreted, generating a plurality of interpretation options for the prediction input by applying the plurality of training models to the prediction input, and outputting the plurality of interpretation options.

Abstract: An evolutionary AutoML framework called LEAF optimizes hyperparameters, network architectures and the size of the network. LEAF makes use of both evolutionary algorithms (EAs) and distributed computing frameworks. A multiobjective evolutionary algorithm is used to maximize the performance and minimize the complexity of the evolved networks simultaneously by calculating the Pareto front given a group of individuals that have been evaluated for multiple objectives.

Abstract: Computer-implemented machines, systems and methods for providing insights about misalignment in a latent space of a machine learning model. A method includes initializing a second weight matrix of a second artificial neural network based on a first weight matrix from a first artificial neural network. The method further includes applying transfer learning between the first artificial neural network and the second artificial neural network. The method further includes comparing the first latent space with the second latent space. The method further includes determining, responsive to the comparing, a first score indicating alignment of the first latent space and the second latent space. The method further includes determining, and responsive to the first score satisfying a threshold, an appropriateness of the machine learning model.

Abstract: The present disclosure relates to a 5G communication system or a 6G communication system for supporting higher data rates beyond a 4G communication system such as long term evolution (LTE). A method of transmitting or receiving a signal by a user equipment (UE) in a mobile communication system is provided. The method may include: identifying a neural network model for transmitting first information to a base station (BS); learning a connection weight of the neural network model using the first information; transmitting, to the base station, second information for updating a weight of a second partial neural network corresponding to the base station based on a result of the learning; and updating a weight of a first partial neural network corresponding to the UE based on the result of the learning.

Abstract: An embodiment of the invention may include a method, computer program product, and system for unified data governance. The embodiment may include populating a context graph with to-be-governed data, by a machine learning framework in communication with a suite of enterprise application servers via one or more connector components. The to-be-governed data is retrieved from the suite of enterprise application servers. The embodiment may include training a plurality of machine learning models using the context graph and the to-be-governed-data based on user-defined parameters. The embodiment may include persisting properties of the plurality of machine learning models back to the context graph.

Abstract: In an approach to automatically downsampling DNA sequence data using variational autoencoders and preserving genomic integrity of an original file embodiments execute, by an encoder, bootstrapping on genomic sequence data to produce resamples. Furthermore, embodiments assess, by the encoder, unrepresentativeness and self-inconsistency of the resamples and selecting a representative resample according to the assessment, and build, by a modified encoder, vector representations from genotype likelihoods based on the selected representative sample. Additionally, embodiments integrate, by an analytics engine, mapping positional information and the genotype likelihoods to identify an optimum vector representation of a resample, and decode, by a modified decoder, the identified optimum vector representation of the resample to obtain a down-sampled read file that resembles and maintains the genomic integrity of the original file.

Abstract: A non-transitory computer-readable storage medium storing a program that causes a computer to execute a process, the process includes obtaining a machine learning model having learned characteristic amounts of a plurality of training data including an objective function; calculating similarities between the characteristic amounts of the plurality of training data by inputting the plurality of training data to the obtained machine learning model; specifying a data group having a high similarity with a desired objective function from the characteristic amounts of the plurality of training data based on distances of the calculated similarities; and acquiring an optimum solution for the desired objective function by using the specified data group.

Abstract: Using a deep learning inference system, respective similarities are measured for each of a set of intermediate representations to input information used as an input to the deep learning inference system. The deep learning inference system includes multiple layers, each layer producing one or more associated intermediate representations. Selection is made of a subset of the set of intermediate representations that are most similar to the input information. Using the selected subset of intermediate representations, a partitioning point is determined in the multiple layers used to partition the multiple layers into two partitions defined so that information leakage for the two partitions will meet a privacy parameter when a first of the two partitions is prevented from leaking information. The partitioning point is output for use in partitioning the multiple layers of the deep learning inference system into the two partitions.