Abstract: A computer-implemented method of training an artificial neural network (ANN) by generating a first learned parameter for use in normalising input data values during a subsequent inference phase of the trained ANN. The method includes, for each of a series of batches of training data values, deriving a batch variance of the batch of training data values and a running variance of all training data values already processed in the training phase; generating an approximation of a current value of the first learned parameter so that a first scaling factor dependent upon the approximation of the first learned parameter and the running variance, is constrained to be equal to a power of two; and normalizing the batch of input data values by a second scaling factor dependent upon the approximation of the current value of the first learned parameter and the batch variance.

Abstract: In a computer-implemented method, an artificial neural network is trained to identify portions of conversation segments within electronic communication documents, wherein an input layer of the artificial neural network includes a plurality of input parameters each corresponding to a different characteristic of text-based content. The method also includes receiving a first electronic communication document that includes first text-based content, and processing the first text-based content using the trained artificial neural network. Processing the first text-based content includes generating one or more position indicators for the first electronic communication document, and the one or more position indicators include one or more segment portion indicators denoting positions of one or more portions of one or more conversation segments within the first electronic communication document.

Abstract: Embodiments generate digital plans for agricultural fields. In an embodiment, a model receives digital inputs including stress risk data, product maturity data, field location data, planting date data, and/or harvest date data. The model mathematically correlates sets of digital inputs with threshold data associated with the stress risk data. The model is used to generate stress risk prediction data for a set of product maturity and field location combinations. In a digital plan, product maturity data or planting date data or harvest date data or field location data can be adjusted based on the stress risk prediction data. A digital plan can be transmitted to a field manager computing device. An agricultural apparatus can be moved in response to a digital plan.

Abstract: Each processor of the SIMD array performs the computations for a respective neuron of a neural network. As part of this computation, each processor of the SIMD array multiplies an input to a weight and accumulates the result for its assigned neuron each (MAC) instruction cycle. A table in a first memory is used to store which input is fed to each processor of the SIMD array. A crossbar is used to route a specific input to each processor each MAC cycle. A second memory is used to provide the appropriate weight to each processor that corresponds the input being routed to that processor.

Abstract: The present disclosure generally relates to techniques for constructing an artificial-intelligence (AI) architecture. The present disclosure relates to techniques for executing the AI architecture to detect whether or not characters in a digital document have been manipulated. The AI architecture can be configured to classify each character in a digital document as manipulated or not manipulated by constructing a graph for each character, generating features for each node of the graph, and inputting a vector representation of the graph into a trained machine-learning model to generate the character classification.

Abstract: Partial differential equations used to simulate physical systems can be solved, in one embodiment, by a solver that has been trained with a set of generative neural networks that operated at different resolutions in a solution space of a domain that defines the physical space of the physical system. The solver can operate in a latent vector space which encodes solutions to the PDE in latent vectors in the latent vector space. The variables of the PDE can be partially decoupled in the latent vector space while the solver operates. The domain can be divided into subdomains that are classified based on their positions in the domain.

Abstract: A neural network includes at least a first network layer that includes a first set of filters and a second network layer that includes a second set of filters. Notably, a filter was removed from the first network layer. A bias associated with a different filter included in the second set of filters compensates for a different bias associated with the filter that was removed from the first network layer.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for obtaining a plurality of model representations of predictive models, each model representation associated with a respective user and expresses a respective predictive model, and selecting a model implementation for each of the model representations based on one or more system usage properties associated with the user associated with the corresponding model representation.

Abstract: Embodiments relate to intelligent entities for providing information service over a network in a telecommunication system. An intelligent element framework manages intelligent entities, which are modular and trained using artificial intelligence or machine learning algorithms to perform prediction or inference for different types of applications. The intelligent entities may communicate with each other via the intelligent element framework. For example, an intelligent entity may generate an output and provide the output for use by one or more other intelligent entities. Thus, the intelligent element framework may distribute portions of tasks for information service across multiple intelligent entities chained together, for example, in a directed graph configuration.

Abstract: Implementations include providing, by the PKG platform, an initial knowledge graph based on user-specific data associated with a user, and a domain-specific knowledge graph, receiving, by the PKG platform, data representative of at least one answer provided from the user to a respective question, providing, by the PKG platform, an expanded knowledge graph based on the initial knowledge graph, the expanded knowledge graph including one or more nodes and respective edges based on the data, generating, by the PKG platform, a weighted knowledge graph based a groundtruth knowledge graph, and a targeted knowledge graph, the groundtruth knowledge graph including one or more true answers, and the targeted knowledge graph including the at least one answer provided from the user, and generating, by the PKG platform, the hyper-personalized knowledge graph (hpKG) based on the weighted knowledge graph, the hpKG being unique to the user within a domain.

Abstract: The disclosure provides a framework or system for learning visual representation using a large set of image/text pairs. The disclosure provides, for example, a method of visual representation learning, a joint representation learning system, and an artificial intelligence (AI) system that employs one or more of the trained models from the method or system. The AI system can be used, for example, in autonomous or semi-autonomous vehicles. In one example, the method of visual representation learning includes: (1) receiving a set of image embeddings from an image representation model and a set of text embeddings from a text representation model, and (2) training, employing mutual information, a critic function by learning relationships between the set of image embeddings and the set of text embeddings.

Abstract: According to one embodiment, an information processing method classifies an instance including a combination of data items of subclasses of either physical world classes describing physical entities or cyber world classes describing concepts. The information processing method comprises the steps of: obtaining first data including the instance; and inferring and determining a subclass the instance belongs to by referring to at least either definition data or log data. The definition data defines the subclasses. The log data includes a set of the first data obtained in the past, each of the first data including the instance with the corresponding subclass defined in the definition data.

Abstract: An information handling system operating a sensor fusion prediction based automatic adjustment system may comprise sensors measuring influencing attributes comprising information handling system operational values, wherein a subset of the influencing attributes influence one of a plurality of system characteristics, and a memory storing definitions of a user behavior characteristic, a performance mapping characteristic, a power status characteristic, a security profile characteristic, and a policy configuration characteristic.

Abstract: There is a need for solutions that generates a temporally dynamic prediction for a particular prediction input. This need can be addressed by, for example, processing the prediction input using each of a plurality of temporally trained machine learning models to generate a corresponding model-specific prediction inference of a plurality of model-specific prediction inferences and processing the plurality of model-specific prediction inferences using an ensemble model to generate the temporally dynamic prediction for the prediction input.

Abstract: A system includes first, second and third input data sets. The first input data set includes demographic information characterizing a patient. The second and third input data sets characterize a healthcare treatment history of the patient. A neural network includes first, second and third neural subnetworks. The first neural subnetwork is configured to process the first input data set to produce a first output data set. The second neural subnetwork is configured to process the second input data set to produce a second output data set. The third neural subnetwork is configured to process the third input data set to produce a third output data set. An autoencoder layer has an input layer comprising the first, second and third output data sets and is configured to process the first, second and third output data sets to produce a secondary output data set.

Abstract: Respective correlation metrics between token groups of a particular text attribute of a data set and a prediction target attribute are computed. Based on the correlation metrics, a predictive token group list is created. For various observation records of the data set, values of a derived categorical attribute corresponding to the particular text attribute are determined based on matches between the particular text attribute value and the predictive token group list. A measure of the predictive utility of the particular text attribute is obtained using correlations between the categorical attribute and the prediction target attribute.

Abstract: A federated machine learning system for training students comprises a first adaptive training system having a first artificial intelligence module for adapting individualized training to a first group of students and for developing a first learning model based on a first set of learning performance metrics. A second adaptive training system provides individualized training to a second group of students and has a data property extraction module for extracting statistical properties from a second set of learning performance metrics for the second group of students. A data simulator module generates simulated performance metrics using extracted statistical properties from the second set of learning performance metrics to thereby generate a second learning model. A federation computing device receives first and second model weights for the first and second learning models and generates or refines a federated model based on the first and second model weights.

Abstract: Disclosed systems, methods, and computer readable media can detect an association between semantic entities and generate semantic information between entities. For example, semantic entities and associated semantic collections present in knowledge bases can be identified. A time period can be determined and divided into time slices. For each time slice, word embeddings for the identified semantic entities can be generated; a first semantic association strength between a first semantic entity input and a second semantic entity input can be determined; and a second semantic association strength between the first semantic entity input and semantic entities associated with a semantic collection that is associated with the second semantic entity can be determined. An output can be provided based on the first and second semantic association strengths.

Abstract: To provide a robust information processing system against attacks by Adversarial Example. A neural network model 608, a latent space database 609 for storing position information in a latent space in which first output vectors, which are output vectors of a predetermined hidden layer included in the neural network model, are embedded concerning input data used for learning of the neural network model, and an inference control unit 606 for making an inference using the neural network model and the latent space database are provided. The inference control unit infers the input data based on the positional relationship between the second output vector, which is an output vector of the predetermined hidden layer concerning input data to be inferred, and the first output vectors in said latent space.

Abstract: An analytics server for scalable predictive analysis for analytics as a software service in multi-tenant environment is provided. The analytics server automatically validates portability of a predictive model from a first tenant to a second tenant by comparing value distribution of parameters between data inputs of the first tenant and the second tenant. The analytics server further automatically detects source data changes over a configurable time horizon as relevant to predictive model inputs, by comparing value distribution of parameters between two data inputs from a same tenant separated by a selected time horizon.