On Adversarial Removal of Hypothesis-only Bias in Natural Language Inference
This addresses bias issues in NLI for researchers and practitioners, but it is incremental as it builds on existing adversarial learning methods.
The paper tackled the problem of hypothesis-only biases in Natural Language Inference datasets by using adversarial learning to encourage models to learn less biased representations, resulting in only small drops in NLI accuracy.
Popular Natural Language Inference (NLI) datasets have been shown to be tainted by hypothesis-only biases. Adversarial learning may help models ignore sensitive biases and spurious correlations in data. We evaluate whether adversarial learning can be used in NLI to encourage models to learn representations free of hypothesis-only biases. Our analyses indicate that the representations learned via adversarial learning may be less biased, with only small drops in NLI accuracy.