# Don't be a Fool: Pooling Strategies in Offensive Language Detection from User-Intended Adversarial Attacks

arXiv: https://arxiv.org/abs/2403.15467 · HTML: https://mlsanity.com/paper/2403.15467/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 16 of its 40 references that are in the map.
- First posted on arXiv 20 Mar 2024.

## Builds on (most cited first, 16 of 16)

- [RoBERTa: A Robustly Optimized BERT Pretraining Approach](https://mlsanity.com/paper/1907.11692/) (2019)
- [Learning Phrase Representations using RNN Encoder-Decoder for Statistical Machine Translation](https://mlsanity.com/paper/1406.1078/) (2014)
- [Unsupervised Cross-lingual Representation Learning at Scale](https://mlsanity.com/paper/1911.02116/) (2019)
- [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://mlsanity.com/paper/2006.03654/) (2020)
- [DeBERTaV3: Improving DeBERTa using ELECTRA-Style Pre-Training with Gradient-Disentangled Embedding Sharing](https://mlsanity.com/paper/2111.09543/) (2021)
- [Is BERT Really Robust? A Strong Baseline for Natural Language Attack on Text Classification and Entailment](https://mlsanity.com/paper/1907.11932/) (2019)
- [TextBugger: Generating Adversarial Text Against Real-world Applications](https://mlsanity.com/paper/1812.05271/) (2018)
- [Predicting the Type and Target of Offensive Posts in Social Media](https://mlsanity.com/paper/1902.09666/) (2019)
- [KLUE: Korean Language Understanding Evaluation](https://mlsanity.com/paper/2105.09680/) (2021)
- [Attacking Neural Text Detectors](https://mlsanity.com/paper/2002.11768/) (2020)
- [BEEP! Korean Corpus of Online News Comments for Toxic Speech Detection](https://mlsanity.com/paper/2005.12503/) (2020)
- [Universal Text Representation from BERT: An Empirical Study](https://mlsanity.com/paper/1910.07973/) (2019)
- [Towards Robust Toxic Content Classification](https://mlsanity.com/paper/1912.06872/) (2019)
- [K-MHaS: A Multi-label Hate Speech Detection Dataset in Korean Online News Comment](https://mlsanity.com/paper/2208.10684/) (2022)
- [Don't Judge a Language Model by Its Last Layer: Contrastive Learning with Layer-Wise Attention Pooling](https://mlsanity.com/paper/2209.05972/) (2022)
- [K-HATERS: A Hate Speech Detection Corpus in Korean with Target-Specific Ratings](https://mlsanity.com/paper/2310.15439/) (2023)

## Built on by (most cited first, 2 of 2)

- [From Ground Trust to Truth: Disparities in Offensive Language Judgments on Contemporary Korean Political Discourse](https://mlsanity.com/paper/2509.14712/) (2025)
- [ContiGuard: A Framework for Continual Toxicity Detection Against Evolving Evasive Perturbations](https://mlsanity.com/paper/2603.14843/) (2026)

## Abstract (arXiv)

Offensive language detection is an important task for filtering out abusive expressions and improving online user experiences. However, malicious users often attempt to avoid filtering systems through the involvement of textual noises. In this paper, we propose these evasions as user-intended adversarial attacks that insert special symbols or leverage the distinctive features of the Korean language. Furthermore, we introduce simple yet effective pooling strategies in a layer-wise manner to defend against the proposed attacks, focusing on the preceding layers not just the last layer to capture both offensiveness and token embeddings. We demonstrate that these pooling strategies are more robust to performance degradation even when the attack rate is increased, without directly training of such patterns. Notably, we found that models pre-trained on clean texts could achieve a comparable performance in detecting attacked offensive language, to models pre-trained on noisy texts by employing these pooling strategies.