FET: Small: Brain-Inspired Hyperdimensional Computing for IoT Applications

University Of California-San Diego, La Jolla CA

Investigators

Abstract

In today's world, technological advances are continually creating more data than what we can cope with. Much of data processing will need to run at least partly on devices at the edge of the internet, such as sensors and smart phones. However, running existing machine learning algorithms on these devices would drain their batteries and be also too slow. Hyper-Dimensional (HD) computing is a class of learning algorithms that is motivated by the observation that the human brain operates on a lot of simple data in parallel. In contrast to today's Deep Neural Networks and other similar algorithms, systems that use HD computing to learn will be able to run at least thousand times more efficiently, can be implemented directly in non-volatile memory, and are natively more secure as they use a large number of bits (~10,000) to encode and process data in parallel. Most importantly, such systems can explain how they made decisions, resulting in sensors and phones that can learn directly from the data they obtain without the need for the cloud at minimum impact to their battery lifetime. This project will develop HD computing software and hardware infrastructure, so that engineers can easily provide HD computing capabilities in their products, and thus benefit from their speed and energy efficiency. The project will support underrepresented minority students including K-12 outreach activities, and disseminate its outcomes and code through open-source efforts. The project seeks to develop: i) novel algorithms supporting key cognitive computations in high-dimensional space including classification, clustering and regression; and ii) novel systems for efficient HD computing on sensors and mobile devices, which cover hardware accelerators such as GPUs, FPGAs and Processor in Memory (PIM), along with software infrastructure to support it. Prototypes will be built and tested in smart homes, and in a large scale sensor network called HPWREN, used for many applications including firefighting, covering 20,000 sq. miles in San Diego area. These demonstrations will show both the quality of the proposed HD algorithms and the efficiency of system designs to address the real-world learning problems. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.