Kernel Approximation using Analog In-Memory Computing

TL;DR

This paper presents a novel kernel approximation method tailored for Analog In-Memory Computing architectures, significantly reducing computational costs while maintaining high accuracy in machine learning tasks.

Contribution

It introduces a hardware-compatible kernel approximation approach that executes in-memory operations, demonstrated on the IBM HERMES AIMC chip, improving efficiency and scalability.

Findings

Less than 1% accuracy drop in kernel ridge classification

Within 1% accuracy on kernelized attention in Transformers

Superior energy efficiency compared to digital accelerators

Abstract

Kernel functions are vital ingredients of several machine learning algorithms, but often incur significant memory and computational costs. We introduce an approach to kernel approximation in machine learning algorithms suitable for mixed-signal Analog In-Memory Computing (AIMC) architectures. Analog In-Memory Kernel Approximation addresses the performance bottlenecks of conventional kernel-based methods by executing most operations in approximate kernel methods directly in memory. The IBM HERMES Project Chip, a state-of-the-art phase-change memory based AIMC chip, is utilized for the hardware demonstration of kernel approximation. Experimental results show that our method maintains high accuracy, with less than a 1% drop in kernel-based ridge classification benchmarks and within 1% accuracy on the Long Range Arena benchmark for kernelized attention in Transformer neural networks. Compared to traditional digital accelerators, our approach is estimated to deliver superior energy efficiency and lower power consumption. These findings highlight the potential of heterogeneous AIMC architectures to enhance the efficiency and scalability of machine learning applications.