A Generative Approach for Mitigating Structural Biases in Natural Language Inference

by   Dimion Asael, et al.

Many natural language inference (NLI) datasets contain biases that allow models to perform well by only using a biased subset of the input, without considering the remainder features. For instance, models are able to make a classification decision by only using the hypothesis, without learning the true relationship between it and the premise. These structural biases lead discriminative models to learn unintended superficial features and to generalize poorly out of the training distribution. In this work, we reformulate the NLI task as a generative task, where a model is conditioned on the biased subset of the input and the label and generates the remaining subset of the input. We show that by imposing a uniform prior, we obtain a provably unbiased model. Through synthetic experiments, we find that this approach is highly robust to large amounts of bias. We then demonstrate empirically on two types of natural bias that this approach leads to fully unbiased models in practice. However, we find that generative models are difficult to train and they generally perform worse than discriminative baselines. We highlight the difficulty of the generative modeling task in the context of NLI as a cause for this worse performance. Finally, by fine-tuning the generative model with a discriminative objective, we reduce the performance gap between the generative model and the discriminative baseline, while allowing for a small amount of bias.



There are no comments yet.


page 1

page 2

page 3

page 4


Don't Take the Premise for Granted: Mitigating Artifacts in Natural Language Inference

Natural Language Inference (NLI) datasets often contain hypothesis-only ...

Discriminatively-Tuned Generative Classifiers for Robust Natural Language Inference

While discriminative neural network classifiers are generally preferred,...

The Gap on GAP: Tackling the Problem of Differing Data Distributions in Bias-Measuring Datasets

Diagnostic datasets that can detect biased models are an important prere...

Unlearn Dataset Bias in Natural Language Inference by Fitting the Residual

Statistical natural language inference (NLI) models are susceptible to l...

Learning from others' mistakes: Avoiding dataset biases without modeling them

State-of-the-art natural language processing (NLP) models often learn to...

simple but effective techniques to reduce biases

There have been several studies recently showing that strong natural lan...

Vision System for AGI: Problems and Directions

What frameworks and architectures are necessary to create a vision syste...
This week in AI

Get the week's most popular data science and artificial intelligence research sent straight to your inbox every Saturday.