Don't Fear the Bit Flips: Optimized Coding Strategies for Binary Classification
This work addresses robustness in binary classification for applications where data is noisy, but it is incremental as it builds on existing channel coding and robustness tools.
The paper tackles the problem of noise-corrupted features in binary classifiers by introducing the same classification probability (SCP) to measure output distortion and proposing optimized coding strategies, including low-complexity SCP estimation and replication error-correcting codes, to maximize SCP with minimal redundancy overhead.
After being trained, classifiers must often operate on data that has been corrupted by noise. In this paper, we consider the impact of such noise on the features of binary classifiers. Inspired by tools for classifier robustness, we introduce the same classification probability (SCP) to measure the resulting distortion on the classifier outputs. We introduce a low-complexity estimate of the SCP based on quantization and polynomial multiplication. We also study channel coding techniques based on replication error-correcting codes. In contrast to the traditional channel coding approach, where error-correction is meant to preserve the data and is agnostic to the application, our schemes specifically aim to maximize the SCP (equivalently minimizing the distortion of the classifier output) for the same redundancy overhead.