Bit Error and Block Error Rate Training for ML-Assisted Communication
This work addresses a key training challenge in ML-assisted communications, offering incremental improvements for coded systems.
The paper tackles the problem of training machine learning-assisted communication systems by showing that the commonly used binary cross-entropy loss is suboptimal in coded systems, and proposes new loss functions and SNR deweighting to minimize block error rates and optimize performance across signal-to-noise ratios, achieving improvements in simulations.
Even though machine learning (ML) techniques are being widely used in communications, the question of how to train communication systems has received surprisingly little attention. In this paper, we show that the commonly used binary cross-entropy (BCE) loss is a sensible choice in uncoded systems, e.g., for training ML-assisted data detectors, but may not be optimal in coded systems. We propose new loss functions targeted at minimizing the block error rate and SNR deweighting, a novel method that trains communication systems for optimal performance over a range of signal-to-noise ratios. The utility of the proposed loss functions as well as of SNR deweighting is shown through simulations in NVIDIA Sionna.