# Interpreting the Robustness of Neural NLP Models to Textual Perturbations

arXiv: https://arxiv.org/abs/2110.07159 · HTML: https://mlsanity.com/paper/2110.07159/

## Key facts

- Cited by 17 papers in the map (data through 14 Oct 2026).
- Builds on 27 of its 46 references that are in the map.
- First posted on arXiv 14 Oct 2021.

## Builds on (most cited first, 27 of 27)

- [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://mlsanity.com/paper/1810.04805/) (2018)
- [RoBERTa: A Robustly Optimized BERT Pretraining Approach](https://mlsanity.com/paper/1907.11692/) (2019)
- [XLNet: Generalized Autoregressive Pretraining for Language Understanding](https://mlsanity.com/paper/1906.08237/) (2019)
- [Character-level Convolutional Networks for Text Classification](https://mlsanity.com/paper/1509.01626/) (2015)
- [SimCSE: Simple Contrastive Learning of Sentence Embeddings](https://mlsanity.com/paper/2104.08821/) (2021)
- [Improved Regularization of Convolutional Neural Networks with Cutout](https://mlsanity.com/paper/1708.04552/) (2017)
- [Beyond Accuracy: Behavioral Testing of NLP models with CheckList](https://mlsanity.com/paper/2005.04118/) (2020)
- [EDA: Easy Data Augmentation Techniques for Boosting Performance on Text Classification Tasks](https://mlsanity.com/paper/1901.11196/) (2019)
- [An Investigation of Why Overparameterization Exacerbates Spurious Correlations](https://mlsanity.com/paper/2005.04345/) (2020)
- [ConSERT: A Contrastive Framework for Self-Supervised Sentence Representation Transfer](https://mlsanity.com/paper/2105.11741/) (2021)
- [Cut, Paste and Learn: Surprisingly Easy Synthesis for Instance Detection](https://mlsanity.com/paper/1708.01642/) (2017)
- [Contextual Augmentation: Data Augmentation by Words with Paradigmatic Relations](https://mlsanity.com/paper/1805.06201/) (2018)
- [Recurrent Neural Network for Text Classification with Multi-Task Learning](https://mlsanity.com/paper/1605.05101/) (2016)
- [An Empirical Study on Robustness to Spurious Correlations using Pre-trained Language Models](https://mlsanity.com/paper/2007.06778/) (2020)
- [Robustness Gym: Unifying the NLP Evaluation Landscape](https://mlsanity.com/paper/2101.04840/) (2021)
- [Text Processing Like Humans Do: Visually Attacking and Shielding NLP Systems](https://mlsanity.com/paper/1903.11508/) (2019)
- [Evaluating the Robustness of Neural Language Models to Input Perturbations](https://mlsanity.com/paper/2108.12237/) (2021)
- [Syntactic Data Augmentation Increases Robustness to Inference Heuristics](https://mlsanity.com/paper/2004.11999/) (2020)
- [Perturbation Sensitivity Analysis to Detect Unintended Model Biases](https://mlsanity.com/paper/1910.04210/) (2019)
- [A Causal View on Robustness of Neural Networks](https://mlsanity.com/paper/2005.01095/) (2020)
- [Robust Open-Vocabulary Translation from Visual Text Representations](https://mlsanity.com/paper/2104.08211/) (2021)
- [Evaluating Robustness to Input Perturbations for Neural Machine Translation](https://mlsanity.com/paper/2005.00580/) (2020)
- [Can neural networks acquire a structural bias from raw linguistic data?](https://mlsanity.com/paper/2007.06761/) (2020)
- [APAC: Augmented PAttern Classification with Neural Networks](https://mlsanity.com/paper/1505.03229/) (2015)
- [Code-Mixing on Sesame Street: Dawn of the Adversarial Polyglots](https://mlsanity.com/paper/2103.09593/) (2021)
- [Improving Neural Machine Translation Robustness via Data Augmentation: Beyond Back Translation](https://mlsanity.com/paper/1910.03009/) (2019)
- [Everything Has a Cause: Leveraging Causal Inference in Legal Text Analysis](https://mlsanity.com/paper/2104.09420/) (2021)

## Built on by (most cited first, 17 of 17)

- [Prompt Injection attack against LLM-integrated Applications](https://mlsanity.com/paper/2306.05499/) (2023)
- [A Close Look into the Calibration of Pre-trained Language Models](https://mlsanity.com/paper/2211.00151/) (2022)
- [Assessing Hidden Risks of LLMs: An Empirical Study on Robustness, Consistency, and Credibility](https://mlsanity.com/paper/2305.10235/) (2023)
- [Sociodemographic Bias in Language Models: A Survey and Forward Path](https://mlsanity.com/paper/2306.08158/) (2023)
- [Robust Natural Language Processing: Recent Advances, Challenges, and Future Directions](https://mlsanity.com/paper/2201.00768/) (2022)
- [Know Thy Strengths: Comprehensive Dialogue State Tracking Diagnostics](https://mlsanity.com/paper/2112.08321/) (2021)
- [An Empirical Study of In-context Learning in LLMs for Machine Translation](https://mlsanity.com/paper/2401.12097/) (2024)
- [Evaluating Robustness of Large Language Models Against Multilingual Typographical Errors](https://mlsanity.com/paper/2510.09536/) (2025)
- [Robustness of Structured Data Extraction from In-plane Rotated Documents using Multi-Modal Large Language Models (LLM)](https://mlsanity.com/paper/2406.10295/) (2024)
- [A Curious Case of Searching for the Correlation between Training Data and Adversarial Robustness of Transformer Textual Models](https://mlsanity.com/paper/2402.11469/) (2024)
- [Stochastic Monkeys at Play: Random Augmentations Cheaply Break LLM Safety Alignment](https://mlsanity.com/paper/2411.02785/) (2024)
- [Greener yet Powerful: Taming Large Code Generation Models with Quantization](https://mlsanity.com/paper/2303.05378/) (2023)
- [JevOut: Natural Context Can Flip Decision Models](https://mlsanity.com/paper/2609.30243/) (2026)
- [Investigating the Impact of Model Instability on Explanations and Uncertainty](https://mlsanity.com/paper/2402.13006/) (2024)
- [How Does Diffusion Influence Pretrained Language Models on Out-of-Distribution Data?](https://mlsanity.com/paper/2307.13949/) (2023)
- [Harmonic LLMs are Trustworthy](https://mlsanity.com/paper/2404.19708/) (2024)
- [CDAE: Enhancing Perturbation Robustness in Pretrained Language Models with Contrastive Denoising](https://mlsanity.com/paper/2607.28236/) (2026)

## Abstract (arXiv)

Modern Natural Language Processing (NLP) models are known to be sensitive to input perturbations and their performance can decrease when applied to real-world, noisy data. However, it is still unclear why models are less robust to some perturbations than others. In this work, we test the hypothesis that the extent to which a model is affected by an unseen textual perturbation (robustness) can be explained by the learnability of the perturbation (defined as how well the model learns to identify the perturbation with a small amount of evidence). We further give a causal justification for the learnability metric. We conduct extensive experiments with four prominent NLP models -- TextRNN, BERT, RoBERTa and XLNet -- over eight types of textual perturbations on three datasets. We show that a model which is better at identifying a perturbation (higher learnability) becomes worse at ignoring such a perturbation at test time (lower robustness), providing empirical support for our hypothesis.