Compression methods improve machine learning accuracy especially in malware detection
An Introduction to Compression-Based Machine Learning
Machine Learning
Summary
Compression techniques that shrink data without losing information can also be used to help computers learn patterns and make predictions. The authors explain how these methods relate to machine learning models and show that tuning design choices in compression-based approaches can boost accuracy. Their experiments found these methods match standard techniques in general and work significantly better for detecting malware. This suggests that combining compression and learning could be a useful approach for some tasks.
What this means in practice
- •For cybersecurity teams: Detect malware more accurately by using compression-based machine learning methods tuned via the proposed design framework.
- •For data scientists: Improve machine learning model accuracy by experimenting with compression-based approaches as alternatives or complements to conventional methods.
Authors
John Hurwitz, Edward Raff, Charles K. Nicholas
Abstract
Any lossless compression algorithm (like gzip) may be converted into a machine learning method, via either Normalized Compression Distance or the Minimum Description Length principle. Any auto-regressive model may be converted into a lossless compression method via entropy coding. This seemingly circular dependence has unrealized potential in modern artificial intelligence and machine learning, and we survey and formalize the various strategies that have been used to leverage compression for machine learning. We introduce and empirically validate a design framework for compression-based ML, finding compression-based methods competitive with conventional baselines and decisively stronger on malware. We find that varying these design choices yields accuracy gains of up to 0.62.