Summary of Genfighter: a Generative and Evolutive Textual Attack Removal, by Md Athikul Islam et al.
GenFighter: A Generative and Evolutive Textual Attack Removal
by Md Athikul Islam, Edoardo Serra, Sushil Jajodia
First submitted to arxiv on: 17 Apr 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Computation and Language (cs.CL)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The paper introduces GenFighter, a novel defense strategy for deep neural networks (DNNs) like Transformer models in natural language processing (NLP), which are vulnerable to adversarial attacks. GenFighter enhances robustness by analyzing the training classification distribution and identifying potentially malicious instances. It transforms these instances into semantically equivalent ones aligned with the training data and uses ensemble techniques for a unified response. The paper demonstrates that GenFighter outperforms state-of-the-art defenses in accuracy under attack and attack success rate metrics, requiring a high number of queries per attack to be effective against NLP adversarial attacks. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper is about making computer programs called deep neural networks more secure. These programs are used for tasks like language translation and can be tricked into giving wrong answers by attackers. The researchers created a new way to make these programs more robust, called GenFighter. It works by looking at the data the program was trained on and identifying any suspicious patterns. Then, it changes those suspicious patterns to look more like the normal patterns in the training data. This makes it harder for attackers to trick the program. The researchers tested this new approach and showed that it is better than other methods at defending against these attacks. |
Keywords
» Artificial intelligence » Classification » Natural language processing » Nlp » Transformer » Translation
