University of California

Mohsen Imani

Tajana Rosing

Farinaz Koushanfar

Sahand Salamat

Mohammad Sadegh Riazi

Yeseong Kim

Behnam Khaleghi

A method of defining an implementation of circuits in a programmable device can be provided by receiving a plurality of specifications for a hyperdimensional (HD) computing machine learning application for execution on a programmable device, determining parameters for a template architecture for HD computing machine learning using the plurality of specifications, the template architecture including an HD hypervector encoder, an HD associative search unit, programmable device pre-defined processing units, and programmable device pre-defined processing elements within the pre-defined processing units, and generating programmable device code configured to specify resources to be allocated within the programmable device using pre-defined circuits defined for use in the programmable device using the determined parameters for the template architecture.

Methods and systems configured to specify resources for hyperdimensional computing implemented in programmable devices using a parameterized template for hyperdimensional computing

Stochastic hyperdimensional arithmetic computing is provided. Hyperdimensional computing (HDC) is a neurally-inspired computation model working based on the observation that the human brain operates on high-dimensional representations of data, called hypervectors. Although HDC is powerful in reasoning and association of the abstract information, it is weak on feature extraction from complex data. Consequently, most existing HDC solutions rely on expensive pre-processing algorithms for feature extraction. This disclosure proposes StocHD, a novel end-to-end hyperdimensional system that supports accurate, efficient, and robust learning over raw data. StocHD expands HDC functionality to the computing area by mathematically defining stochastic arithmetic over HDC hypervectors. StocHD enables an entire learning application (including feature extractor) to process using HDC data representation, enabling uniform, efficient, robust, and highly parallel computation. This disclosure further provides a novel fully digital and scalable processing in-memory (PIM) architecture that exploits the HDC memory-centric nature to support extensively parallel computation.

Stochastic hyperdimensional arithmetic computing

Disclosed is a network-based hyperdimensional system having an encoder configured to receive input data and encode the input data using hyperdimensional computing to generate a hypervector having encoded data bits that represent the input data. The network-based hyperdimensional system further includes a decoder configured to receive the encoded data bits, decode the encoded data bits, and reconstruct the input data from the decoded data bits. In some embodiments, the encoder is configured for direct hyperdimensional learning on transmitted data with no need for data decoding by the decoder.

Network-based hyperdimensional system

A computing system can include a plurality of clients located outside a cloud-based computing environment, where each of the clients may be configured to encode respective original data with a respective unique secret key to generate data hypervectors that encode the original data. A collaborative machine learning system can operate in the cloud-based computing environment and can be operatively coupled to the plurality of clients, where the collaborative machine learning system can be configured to operate on the data hypervectors that encode the original data to train a machine learning model operated by the collaborative machine learning system or to generate an inference from the machine learning model.

Systems, circuits and computer program products providing a framework for secured collaborative training using hyper-dimensional vector based data encoding/decoding and related methods

Growing community of inventors

San Diego, CA, United States of America

Mohsen Imani