Toward Storage-Aware Learning with Compressed Data An Empirical Exploratory Study on JPEG

2025-08-18

AI Summary

In device-edge continual learning, a fundamental tension exists between stringent storage constraints and the degradation of data quality due to compression. Method: This paper systematically investigates the data compression trade-offs inherent in storage-aware learning, revealing significant heterogeneity in sample-wise sensitivity to compression—rendering uniform or random compression/selection strategies suboptimal. We formally characterize, for the first time, the tripartite trade-off among compression intensity, storage budget, and model performance, and propose a sample-level adaptive compression framework. Contribution/Results: Empirical evaluation using JPEG compression demonstrates that, under identical storage budgets, our method improves model accuracy by 3.2–5.7 percentage points over baseline approaches. This work establishes a novel paradigm and scalable technical pathway for building efficient and robust edge-intelligent learning systems.

Technology Category

• Machine Learning: Learning on the Edge & Model Compression
• Data Mining & Knowledge Management: Data Compression
• Search and Optimization: Learning to Search

Application Category

• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

On-device machine learning is often constrained by limited storage, particularly in continuous data collection scenarios. This paper presents an empirical study on storage-aware learning, focusing on the trade-off between data quantity and quality via compression. We demonstrate that naive strategies, such as uniform data dropping or one-size-fits-all compression, are suboptimal. Our findings further reveal that data samples exhibit varying sensitivities to compression, supporting the feasibility of a sample-wise adaptive compression strategy. These insights provide a foundation for developing a new class of storage-aware learning systems. The primary contribution of this work is the systematic characterization of this under-explored challenge, offering valuable insights that advance the understanding of storage-aware learning.

Problem

• Explores storage-aware learning with compressed data
• Studies trade-off between data quantity and quality
• Proposes sample-wise adaptive compression strategy

Innovation

• Sample-wise adaptive compression strategy
• Trade-off between data quantity and quality
• Storage-aware learning systems development