Challenge: Existing adversarial training methods require multi-step gradient ascents or word substitutions to obtain adversarials, which impairs the effectiveness of adversariarial training.
Approach: They propose a procedure for instead adversarial training with only clean data that estimates the adversarials loss by perturbing the input data’s probability distribution rather than their embeddings.
Outcome: The proposed procedure reduces time consumption by up to 70% compared to current best-performing adversarial training methods.

Similar Papers

Flooding-X: Improving BERT’s Resistance to Adversarial Attacks via Loss-Restricted Fine-Tuning (2022.acl-long)

Copied to clipboard

Challenge: Existing approaches to generating adversarial perturbations scale up the cost of training computational complexity by the number of gradient steps it takes to obtain the adversarials.
Approach: They propose a flood method which aims at better generalization and a criterion to bring hyper-parameter-dependent flooding into effect with a narrowed-down search space by measuring how the gradient steps taken within one epoch affect the loss of each batch.
Outcome: The proposed method improves BERT’s resistance to textual adversarial attacks by a large margin and achieves state-of-the-art robust accuracy on various text classification and GLUE tasks.
SemRoDe: Macro Adversarial Training to Learn Representations that are Robust to Word-Level Attacks (2024.naacl-long)

Copied to clipboard

Challenge: Existing approaches to defend against word-level attacks have been limited.
Approach: They propose a new approach called Semantic Robust Defence to enhance the robustness of language models by aligning the domains with a distance-based objective.
Outcome: The proposed approach can be generalized across word embeddings, even when they share minimal overlap at both vocabulary and word-substitution levels.
Don’t Take the Easy Way Out: Ensemble Based Methods for Avoiding Known Dataset Biases (D19-1)

Copied to clipboard

Challenge: Recent advances in neural models exploit dataset-specific patterns that do not generalize well to out-of-domain or adversarial settings.
Approach: They propose to train a model to be more robust to domain shift if it has prior knowledge of dataset biases.
Outcome: The proposed model can be more robust to domain shift if it has prior knowledge of dataset biases.
DA3: A Distribution-Aware Adversarial Attack against Language Models (2024.emnlp-main)

Copied to clipboard

Challenge: Recent attacks have shown that adversarial examples have a different data distribution than the original examples, reducing their effectiveness under detection methods.
Approach: They propose a distribution-aware adversarial attack method that considers the distribution shifts of adversarials to improve attacks’ effectiveness under detection methods.
Outcome: The proposed method improves the effectiveness of adversarial examples under detection methods and integrates both ASR and detectability.
Domain Adversarial Fine-Tuning as an Effective Regularizer (2020.findings-emnlp)

Copied to clipboard

Challenge: Existing fine-tuning techniques can degrade general-domain representations . however, fine-timing can lead to catastrophic forgetting of knowledge .
Approach: They propose a new regularization technique that complements the task-specific loss used during fine-tuning with an adversarial objective.
Outcome: Empirical results show that AFTER improves performance on various natural language understanding tasks compared to standard fine-tuning.
Improving Gradient-based Adversarial Training for Text Classification by Contrastive Learning and Auto-Encoder (2021.findings-acl)

Copied to clipboard

Challenge: Recent work has shown that models can be easily fooled by intentionally designed adversarial examples.
Approach: They propose two efficient approaches for generating adversarial perturbations on embeddings and propose two new approaches to help model learn adversarials more efficiently.
Outcome: The proposed approaches outperform strong baselines on various text classification datasets and the model's performance drops less under adversarial attack.
Adversarial Preference Learning for Robust LLM Alignment (2025.findings-acl)

Copied to clipboard

Challenge: Modern language models rely on Reinforcement Learning from Human Feedback (RLHF) to encourage safe behaviors, but they remain vulnerable to adversarial attacks due to three key limitations: (1) the inefficiency and high cost of human annotation; (2) the vast diversity of potential adversarials; and (3) the risk of feedback bias and reward hacking.
Approach: They propose an iterative adversarial training method that incorporates three key innovations to address these challenges.
Outcome: Experiments on Mistral-7B-Instruct-v0.3 show that the proposed method significantly enhances robustness and reduces harmful outputs from 5.88% to 0.43%.
BERT-ATTACK: Adversarial Attack Against BERT Using BERT (2020.emnlp-main)

Copied to clipboard

Challenge: Current approaches to generate adversarial samples for discrete data are heuristic replacement strategies that are difficult to implement in continuous data.
Approach: They propose a method to generate adversarial samples using pre-trained masked language models using BERT.
Outcome: The proposed method outperforms state-of-the-art methods in success rate and perturb percentage while remaining fluent and semantically preserved.
Generative Adversarial Training with Perturbed Token Detection for Model Robustness (2023.emnlp-main)

Copied to clipboard

Challenge: Existing adversarial training methods use discrete tokens to deceive models . current approaches use embeddings, whereas actual text-based training uses discrete text tokens.
Approach: They propose a framework that integrates gradient-based learning, adversarial example generation and perturbed token detection to enhance adversariarial robustness.
Outcome: The proposed framework surpasses the state-of-the-art results of ChatGPT by 10% in average accuracy.
Step by Step Loss Goes Very Far: Multi-Step Quantization for Adversarial Text Attacks (2023.eacl-main)

Copied to clipboard

Challenge: Existing gradient-based attacks quantize all tokens in a text at once, which creates a significant gap between adversarial loss for continuous and discrete text representations.
Approach: They propose a gradient-based attack that quantizes tokens one by one and reoptimizes adversarial example after each quantization.
Outcome: The proposed method outperforms other approaches on various natural language processing tasks.

What is GenGO?

GenGO is an NLP powered publication search system. It currenctly indexes 30k+ papers from ACL Anthology, and implements multi-aspect summarization, semantic search, and more!

Information

About
Limitations