Stochastic Multivariate Universal-Radix Finite-State Machine: a Theoretically and Practically Elegant Nonlinear Function Approximator

TL;DR

This paper introduces SMURF, a stochastic finite-state machine that efficiently approximates complex nonlinear functions with significantly reduced hardware resources, enhancing neural network performance.

Contribution

The paper presents the first stochastic multivariate universal-radix FSM architecture for nonlinear function approximation, combining theoretical analysis and practical hardware efficiency.

Findings

SMURF achieves 16.07% area and 14.45% power of Taylor-series methods.

SMURF uses only 2.22% area of LUT schemes.

High accuracy in multivariate nonlinear approximation.

Abstract

Nonlinearities are crucial for capturing complex input-output relationships especially in deep neural networks. However, nonlinear functions often incur various hardware and compute overheads. Meanwhile, stochastic computing (SC) has emerged as a promising approach to tackle this challenge by trading output precision for hardware simplicity. To this end, this paper proposes a first-of-its-kind stochastic multivariate universal-radix finite-state machine (SMURF) that harnesses SC for hardware-simplistic multivariate nonlinear function generation at high accuracy. We present the finite-state machine (FSM) architecture for SMURF, as well as analytical derivations of sampling gate coefficients for accurately approximating generic nonlinear functions. Experiments demonstrate the superiority of SMURF, requiring only 16.07% area and 14.45% power consumption of Taylor-series approximation, and merely 2.22% area of look-up table (LUT) schemes.