Gender Bias Amplification During Speed-Quality Optimization in Neural Machine Translation
This addresses bias amplification in machine translation for users and developers, highlighting a critical issue in speed-quality trade-offs.
The study investigated whether gender bias is amplified when neural machine translation models are optimized for speed, finding that while overall BLEU scores degrade minimally, gendered noun translation performance degrades much faster.
Is bias amplified when neural machine translation (NMT) models are optimized for speed and evaluated on generic test sets using BLEU? We investigate architectures and techniques commonly used to speed up decoding in Transformer-based models, such as greedy search, quantization, average attention networks (AANs) and shallow decoder models and show their effect on gendered noun translation. We construct a new gender bias test set, SimpleGEN, based on gendered noun phrases in which there is a single, unambiguous, correct answer. While we find minimal overall BLEU degradation as we apply speed optimizations, we observe that gendered noun translation performance degrades at a much faster rate.