Abstract: A quantum prediction AI system includes a quantum prediction circuit adapted to receive an input vector representing a subset of a time-sequential sequence; encode the input vector as a corresponding qubit register; apply a trained quantum circuit to the qubit register; and measure one or more qubits output from the quantum prediction circuit to infer a next data point in the series following the subset represented by the input vector.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training a machine learning model to generate embeddings of inputs to the machine learning model, the machine learning model having an encoder that generates the embeddings from the inputs and a decoder that generates outputs from the generated embeddings, wherein the embedding is partitioned into a sequence of embedding partitions that each includes one or more dimensions of the embedding, the operations comprising: for a first embedding partition in the sequence of embedding partitions: performing initial training to train the encoder and a decoder replica corresponding to the first embedding partition; for each particular embedding partition that is after the first embedding partition in the sequence of embedding partitions: performing incremental training to train the encoder and a decoder replica corresponding to the particular partition.

Abstract: Techniques herein train a multilayer perceptron, sparsify edges of a graph such as the perceptron, and store edges and vertices of the graph. Each edge has weight. A computer sparsifies perceptron edges. The computer performs a forward-backward pass on the perceptron to calculate a sparse Hessian matrix. Based on that Hessian, the computer performs quasi-Newton perceptron optimization. The computer repeats this until convergence. The computer stores edges in an array and vertices in another array. Each edge has weight and input and output indices. Each vertex has input and output indices. The computer inserts each edge into an input linked list based on its weight. Each link of the input linked list has the next input index of an edge. The computer inserts each edge into an output linked list based on its weight. Each link of the output linked list comprises the next output index of an edge.

Abstract: An analog neuromorphic circuit is disclosed having resistive memories that provide a resistance to an input voltage signal as the input voltage signal propagates through the resistive memories generating a first output voltage signal and to provide a resistance to a first error signal that propagates through the resistive memories generating a second output voltage signal. A comparator generates the first error signal that is representative of a difference between the first output voltage signal and the desired output signal and generates the first error signal so that the first error signal propagates back through the plurality of resistive memories. A resistance adjuster adjusts a resistance value associated with each resistive memory based on the first error signal and the second output voltage signal to decrease the difference between the first output voltage signal and the desired output signal.

Abstract: Devices and methods for systolically processing data according to a neural network. In one aspect, a first arrangement of processing units includes at least first, second, third, and fourth processing units. The first and second processing units are connected to systolically pulse data to one another, and the third and fourth processing units are connected to systolically pulse data to one another. A second arrangement of processing units includes at least fifth, sixth, seventh, and eighth processing units. The fifth and sixth processing units are connected to systolically pulse data to one another, and the seventh and eighth processing units are connected to systolically pulse data to one another. The second processing unit is configured to systolically pulse data to the seventh processing unit along a first interconnect and the third processing unit is configured to systolically pulse data to the sixth processing unit along a second interconnect.

Abstract: Methods, systems, and computer-readable storage media for receiving a dataset, the dataset including a plurality of data values, clustering data values of the plurality of data values into a plurality of input feature clusters in input feature space, training a local multi-task Gaussian process (MTGP) for each input feature cluster to provide optimized hyper-parameters in hyper-parameter space, an optimized hyper-parameter being provided for each input feature cluster, merging data values based on the optimized hyper-parameters, and distances between hyper-parameter clusters in the hyper-parameter space to provide a plurality of merged data values, and providing a LL-MTGP model based on the merged data values.

Abstract: A system for monitoring an environment may include an input device for monitoring and capturing pattern-based states of a model of the environment. The system may also include a thalamobot embodied in at least a first processor in communication with the input device. The thalamobot may include at least one filter for monitoring captured data from the input device and for identifying at least one state change within the captured data. The system may also include at least one critic and/or at least one recognition system. The at least one filter forwards said at least one state change to the critic and/or recognition system. Novel schemes are introduced to allow processors to interconnect themselves into brain-like structures that contemplate both the environment and the model thereof, unifying disparate data into discoveries. The significance of such discoveries is recognized either through neural activation patterns or the topologies of interconnecting neural modules.

Abstract: One embodiment of an accelerator includes a computing unit; a first memory bank for storing input activations and a second memory bank for storing parameters used in performing computations, the second memory bank configured to store a sufficient amount of the neural network parameters on the computing unit to allow for latency below a specified level with throughput above a specified level. The computing unit includes at least one cell comprising at least one multiply accumulate (“MAC”) operator that receives parameters from the second memory bank and performs computations. The computing unit further includes a first traversal unit that provides a control signal to the first memory bank to cause an input activation to be provided to a data bus accessible by the MAC operator. The computing unit performs computations associated with at least one element of a data array, the one or more computations performed by the MAC operator.

Abstract: A method and associated systems for a database-management system with artificially intelligent database administration. The DBMS manages one or more databases, each of which is monitored by sensors that detect conditions indicative of database performance. An operational engine receives input from the sensors, translates it into a form understandable by an artificially intelligent decision engine, and forwards the translated input to the decision engine. The decision engine uses preloaded knowledge elements stored in a knowledgebase to infer whether the sensor input identifies an issue that can only be resolved by a database-administration activity. If so, the decision engine attempts to select a best solution, optionally seeks confirmation of its selection from an outside source, and directs the operational engine to implement the selected solution.

Abstract: An exemplary embodiment may present a behavior modeling architecture that is intended to assist in handling, modelling, predicting and verifying the behavior of machine learning models to assure the safety of such systems meets the required specifications and adapt such architecture according to the execution sequences of the behavioral model. An embodiment may enable conditions in a behavioral model to be integrated in the execution sequence of behavioral modeling in order to monitor the probability likelihoods of certain paths in a system. An embodiment allows for real-time monitoring during training and prediction of machine learning models. Conditions may also be utilized to trigger system-knowledge injection in a white-box model in order to maintain the behavior of a system within defined boundaries. An embodiment further enables additional formal verification constraints to be set on the output or internal parts of white-box models.

Abstract: Aspects of the present disclosure provide techniques for machine learning and rules integration. Embodiments include receiving input values corresponding to a subset of a set of input variables associated with an automated determination. Embodiments include generating a directed acyclic graph (DAG) representing a set of constraints corresponding to the set of input variables. The set of constraints relate to one or more machine learning models and one or more rules. Embodiments include receiving one or more outputs from the one or more machine learning models based on one or more of the input values. Embodiments include determining outcomes for the one or more rules based on at least one of the input values. Embodiments include populating the DAG based on the input values, the one or more outputs, and the outcomes. Embodiments include making the automated determination based on logic represented by the DAG.

Abstract: Computer-implemented, machine-learning systems and methods relate to a neural network having at least two subnetworks, i.e., a first subnetwork and a second subnetwork. The systems and methods estimate the partial derivative(s) of an objective with respect to (i) an output activation of a node in first subnetwork, (ii) the input to the node, and/or (iii) the connection weights to the node. The estimated partial derivative(s) are stored in a data store and provided as input to the second subnetwork. Because the estimated partial derivative(s) are persisted in a data store, the second subnetwork has access to them even after the second subnetwork has gone through subsequent training iterations. Using this information, subnetwork 160 can compute classifications and regression functions that can help, for example, in the training of the first subnetwork.

Abstract: A method for determining a risk of decompensated heart failure in a user includes receiving a first set of data that is fixed with respect to time. A machine-learning model generates one or more initial risk factors based on the first set of data. A second set of data for the user that dynamically updates over time is received from a wearable cardiovascular physiology monitor. The machine-learning model is used to generate dynamic data classifiers based on the one or more initial risk factors. Aggregate risk scores for the user are then indicated based on an evaluation of the second set of data against the dynamic data classifiers. In this way, static electronic medical records may be combined with dynamic, real-time data from wearable cardiovascular physiology monitors to provide an accurate and continuously updating risk of decompensated heart failure for a user.

Abstract: A system for determining risky events includes an input interface and a processor. The input interface is for receiving sensor data on environmental conditions. The processor is for determining whether the environmental conditions indicate an increase in event probability and, in the event that environmental conditions indicate the increase in the event probability, adjusting an event detection threshold.

Abstract: Methods, systems and computer program products are provided for cross-media recommendation by store a plurality of taste profiles corresponding to a first domain and a plurality of media item vectors corresponding to a second domain. An evaluation taste profile in the first domain is applied to a plurality of models that have been generated based on relationship among the plurality of taste profiles and the plurality of media item vectors, and obtain a plurality of resulting codes corresponding to at least one of the plurality of media item vectors in the second domain.

Abstract: A system and method for machine understanding, using program induction, includes a visual cognitive computer including a set of components designed to execute predetermined primitive functions. The method includes determining programs using a program induction engine that interfaces with the visual cognitive computer to discover programs using the predetermined primitive functions and/or executes the discovered programs based on an input.

Abstract: A learning-type signal separation method performed using a model formulation unit, which performs learning processing based on a training-use signal including a specific component, and a training-use signal not including the specific component, the training-use signals including a common characteristic. The learning-type signal separation method includes: generating learned data by causing the model formulation unit to perform learning processing based on the training-use signal and information indicating whether or not the specific component is included in the training-use signal, to generate a data series signal in which the specific component has been separated and removed from a data series of the training-use signal; acquiring an arbitrary signal including the common characteristic; and generating, based on the acquired arbitrary signal and the generated learned data, a data series signal in which the specific component has been separated and removed from a data series of the arbitrary signal.

Abstract: According to an embodiment, a neural network apparatus includes cores, routers, a tree path, and a short-cut path. The cores are provided according to leaves in a tree structure, each core serving as a circuit that performs calculation or processing for part of elements of the neural network. The routers are provided according to nodes other than the leaves in the tree structure. The tree path connects the cores and the routers such that data is transferred along the tree structure. The short-cut path connects part of the routers such that data is transferred on a route differing from the tree path. The routers transmit data output from each core to any of the cores serving as a transmission destination on one of routes in the tree path and the short-cut path such that the calculation or the processing is performed according to a structure of the neural network.

Abstract: A method, apparatus and product includes obtaining a logical representation of a quantum circuit that is implementable by a plurality of alternative physical representations of the quantum circuit, each of which implementing the logical representation with a different error correction scheme and defining error correction schemes for the quantum circuit. The defining error correction schemes includes implementing a search algorithm on the alternative physical representations, wherein the search algorithm is configured to search for a physical representation of the quantum circuit with an assignment of a plurality of physical qubits to a plurality of logical qubits that is defined in view of a quality score. A quality metric used to compute the quality score is monotonically correlated to error rates of logical output qubits of the quantum circuit when implementing each alternative physical representation. The assignment is utilized to define the error correction schemes for the quantum circuit.

Abstract: Approaches, techniques, and mechanisms are disclosed for generating, enhancing, applying and updating knowledge neurons for providing decision making information to a wide variety of client applications. Domain keywords for knowledge domains are generated from domain data of selected domain data sources, along with keyword values for the domain keywords, and are used to generate knowledge artifacts for inclusion in knowledge neurons. These knowledge neurons may be enhanced by domain knowledge data sets found in various data sources and used to generate neural responses to neural queries received from the client applications. Neural feedbacks may be used to update and/or generate knowledge neurons.