Towards Robustifying NLI Models Against Lexical Dataset Biases

By Xiang Zhou, Mohit Bansal. 2020

62 citations recorded by the source corpus.

While deep learning models are making fast progress on the task of Natural Language Inference, recent studies have also shown that these models achieve high accuracy by exploiting several dataset biases, and without deep understanding of the language semantics. Using contradiction-word bias and word-overlapping bias as our two bias examples, this paper explores both data-level and model-level debiasing methods to robustify models against lexical dataset biases. First, we debias the dataset through data augmentation and enhancement, but show that the model bias cannot be fully removed via this method. Next, we also compare two ways of directly debiasing the model without knowing what the dataset biases are in advance. The first approach aims to remove the label bias at the embedding level. The second approach employs a bag-of-words sub-model to capture the features that are likely to exploit the bias and prevents the original model from learning these biased features by forcing orthogonality between these two sub-models. We performed evaluations on new balanced datasets extracted from the original MNLI dataset as well as the NLI stress tests, and show that the orthogonality approach is better at debiasing the model while maintaining competitive overall accuracy.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Natural Language Processing, Safety, Robustness & Interpretability

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

In Natural Language Processing

In Safety, Robustness & Interpretability

Citations in the corpus

Flexible Generation of Natural Language Deductions (2021, background): “We hypothesize that while this model may assign higher likelihood to valid conclusions than a general language model would, it will place much more probability mass on reemitting premise statements due to the fact that high word overlap tends to be a common feature of RTE examples labeled as ‘entail”
An Empirical Study on Robustness to Spurious Correlations using Pre-trained Language Models (2020, background): “Serveral recent work proposes new learning algorithms that are robust to spurious correlations in NLI datasets (He et al., 2019; Clark et al., 2019; Yaghoobzadeh et al., 2019; Zhou and Bansal, 2020; Sagawa et al., 2020; Mahabadi et al., 2020; Utama et al., 2020).”
Generating Data to Mitigate Spurious Correlations in Natural Language Inference Datasets (2022, methodology): “Zhou and Bansal (2020) use HEX projection to project the representation to the space orthogonal to the biased features to debias the model.”
Evaluating Factuality in Generation with Dependency-level Entailment (2020, background): “These errors are tougher for the MNLI model to catch, which evaluates the whole sentence and is prone to lexical overlap biases (Zhou and Bansal, 2020).”
Evaluating Gender Bias in Natural Language Inference (2021, background): “This has prompted a large area of research attempting to evaluate and mitigate them, either through removal of bias introduction in dataset level (Barbosa & Chen, 2019), or through model architecture (Gonen & Goldberg, 2019), or both (Zhou & Bansal, 2020).”