Various embodiments are generally directed to techniques for determining artificial neural network topologies, such as by utilizing probabilistic graphical models, for instance. Some embodiments are particularly related to determining neural network topologies by bootstrapping a graph, such as a probabilistic graphical model, into a multi-graphical model, or graphical model tree. Various embodiments may include logic to determine a collection of sample sets from a dataset. In various such embodiments, each sample set may be drawn randomly for the dataset with replacement between drawings. In some embodiments, logic may partition a graph into multiple subgraph sets based on each of the sample sets. In several embodiments, the multiple subgraph sets may be scored, such as with Bayesian statistics, and selected amongst as part of determining a topology for a neural network.

A system and method for generating data visualizations. The method includes generating an enriched data layer based on a plurality of knowledge graphs, the plurality of knowledge graphs including a plurality of first nodes, the enriched data layer including a plurality of second nodes, wherein each of the plurality of second nodes is connected via an edge to at least one of the plurality of first nodes; and generating a data visualization based on the enriched data layer and a request for data, wherein the request for data indicates a type of data corresponding to at least one of the plurality of second nodes, wherein the data visualization is generated using data represented by at least one of the plurality of first nodes connected to the at least one of the plurality of second nodes.

A method and apparatus are provided that analyzing sequence-to-sequence data, such as sequence-to-sequence speech data or sequence-to-sequence machine translation data for example, by minimum Bayes risk (MBR) training a sequence-to-sequence model and within introduction of applications of softmax smoothing to an N-best generation of the MBR training of the sequence-to-sequence model.

Embodiments for automatic visualization and explanation of feature learning output for predictive modeling in a computing environment by a processor. A degree of importance score may be assigned to one or more features from a relational database according to the machine learning model. A visualization graph of one or more join paths and the one or more features with the degree of importance score to predict a target variable may be generated.

There is a need for solutions that classification solutions in hierarchical prediction domains. This need can be addressed by, for example, performing one or more online machine learning, co-occurrence analysis machine learning, structured fusion machine learning, and unstructured fusion machine learning. In one example, structured predictions inputs are processed in accordance with an online machine learning analysis to generate structurally hierarchical predictions and in accordance with a co-occurrence analysis machine learning analysis to generate structurally non-hierarchical predictions. Then, the structurally hierarchical predictions and the structurally non-hierarchical predictions in accordance with processed by a structured fusion model to generate structure-based predictions. Afterward, the structure-based predictions and non-structure-based predictions are processed in accordance with an unstructured fusion model to generate one or more unstructured-fused predictions.

Neural network-based categorization can be improved by incorporating graph neural networks that operate on a graph representing the taxonomy of the categories into which a given input is to be categorized by the neural network based-categorization. The output of a graph neural network, operating on a graph representing the taxonomy of categories, can be combined with the output of a neural network operating upon the input to be categorized, such as through an interaction of multidimensional output data, such as a dot product of output vectors. In such a manner, information conveying the explicit relationships between categories, as defined by the taxonomy, can be incorporated into the categorization. To recapture information, incorporate new information, or reemphasize information a second neural network can also operate upon the input to be categorized, with the output of such a second neural network being merged with the output of the interaction.

Methods, apparatus, systems, and articles of manufacture are disclosed to decode receipts based on neural graph architecture. An example apparatus for decoding receipts includes, vertex feature representation circuitry to extract features from optical-character-recognition (OCR) words, polar coordinate circuitry to: calculate polar coordinates of the OCR words based on respective ones of the extracted features, graph neural network circuitry to generate an adjacency matrix based on the extracted features, post-processing circuitry to traverse the adjacency matrix to generate cliques of OCR processed words, and output circuitry to generate lines of text based on the cliques of OCR processed words.

Systems and methods for determining whether an electronic document constitutes vouching evidence is provided. The system may receive ERP item data and generate hypothesis data based thereon, and may receive electronic document data and extract ERP information therefrom. The system may then apply one or more models to compare the hypothesis data to the extracted ERP information to determine whether the electronic document constitutes vouching evidence for the ERP item. Systems and methods for verifying an assertion against a source document are provided. The system may receive first data indicating an unverified assertion and second data comprising a plurality of source documents. The system may apply one or more extraction models to extract a set of key data from the plurality of source documents and may apply one or more matching models to compare the first data to the set of key data to determine whether vouching criteria are met.

The disclosure relates to method and system for on-device inference in a deep neural network (DNN). The method comprises: determining whether one or more layers of the DNN satisfy one of a first, a second and a third condition, the one or more layers including one or more convolution layers and one or more resampling layers; performing the on-device inference based on the determination, wherein performing the on-device inference comprises at least one of: optimizing the one or more convolution layers in the one or more parallel branches based on the one or more layers of the DNN satisfying the first condition, optimizing the at least one of the resampling layers based on the one or more layers of the DNN satisfying the second condition, and modifying operation of the at least one of the resampling layers based on the one or more layers of the DNN satisfying the third condition.

A method of detecting gaits of an individual with a sensor worn by the individual. The sensor includes an accelerometer and a processing unit. The method includes obtaining an signal representing one or more sensor acceleration values; sampling the signal to obtain a sampled signal; segmenting the sampled signal into windows to obtain a segmented acceleration signal; extracting a feature set from the segmented acceleration signal; determining a probability value, for a respective window, n, where n is a positive integer greater than zero, the probability value giving an estimated probability value of gait occurrence for the individual during the respective window; modifying the estimated probability value by using a histogram of previously detected gait durations to obtain a modified probability value; and determining, based on the modified probability value, and by using a determination threshold whether or not the respective window represents gait occurrence.