Understanding The Effectiveness of Lossy Compression in Machine Learning Training Sets

TL;DR

This paper systematically evaluates 17 lossy data compression methods across 7 ML/AI applications, demonstrating significant compression ratios with minimal quality loss, and provides insights for future compressor design.

Contribution

It introduces a comprehensive evaluation methodology and offers the first extensive comparison of lossy compression effects on diverse ML/AI tasks.

Findings

Achieves 50-100x compression with ≤1% quality loss

Modern lossy methods outperform traditional approaches

Provides guidelines for designing ML/AI-friendly compressors

Abstract

Learning and Artificial Intelligence (ML/AI) techniques have become increasingly prevalent in high performance computing (HPC). However, these methods depend on vast volumes of floating point data for training and validation which need methods to share the data on a wide area network (WAN) or to transfer it from edge devices to data centers. Data compression can be a solution to these problems, but an in-depth understanding of how lossy compression affects model quality is needed. Prior work largely considers a single application or compression method. We designed a systematic methodology for evaluating data reduction techniques for ML/AI, and we use it to perform a very comprehensive evaluation with 17 data reduction methods on 7 ML/AI applications to show modern lossy compression methods can achieve a 50-100x compression ratio improvement for a 1% or less loss in quality. We identify critical insights that guide the future use and design of lossy compressors for ML/AI.