Adversarial Semantic Collisions
Abstract
We study semantic collisions: texts that are semantically unrelated but judged as similar by NLP models. We develop gradient-based approaches for generating semantic collisions and demonstrate that state-of-the-art models for many tasks which rely on analyzing the meaning and similarity of texts-- including paraphrase identification, document retrieval, response suggestion, and extractive summarization-- are vulnerable to semantic collisions. For example, given a target query, inserting a crafted collision into an irrelevant document can shift its retrieval rank from 1000 to top 3. We show how to generate semantic collisions that evade perplexity-based filtering and discuss other potential mitigations. Our code is available at https://github.com/csong27/collision-bert.
Cite
@article{arxiv.2011.04743,
title = {Adversarial Semantic Collisions},
author = {Congzheng Song and Alexander M. Rush and Vitaly Shmatikov},
journal= {arXiv preprint arXiv:2011.04743},
year = {2020}
}