Hyperdimensional Vector Tsetlin Machines with Applications to Sequence Learning and Generation

TL;DR

This paper introduces a two-layered hyperdimensional vector Tsetlin machine model that efficiently learns and generates sequential data, combining hyperdimensional computing with Tsetlin machine interpretability for improved sequence forecasting and classification.

Contribution

The paper presents a novel two-layered model integrating hyperdimensional vector computing with Tsetlin machines, enhancing sequence learning and generation capabilities.

Findings

Competitive performance on UCR Time Series Archive

Efficient sequence forecasting and generation

Interpretable model structure

Abstract

We construct a two-layered model for learning and generating sequential data that is both computationally fast and competitive with vanilla Tsetlin machines, adding numerous advantages. Through the use of hyperdimensional vector computing (HVC) algebras and Tsetlin machine clause structures, we demonstrate that the combination of both inherits the generality of data encoding and decoding of HVC with the fast interpretable nature of Tsetlin machines to yield a powerful machine learning model. We apply the approach in two areas, namely in forecasting, generating new sequences, and classification. For the latter, we derive results for the entire UCR Time Series Archive and compare with the standard benchmarks to see how well the method competes in time series classification.