# Logically Consistent Adversarial Attacks for Soft Theorem Provers

arXiv: https://arxiv.org/abs/2205.00047 · HTML: https://mlsanity.com/paper/2205.00047/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 23 of its 33 references that are in the map.
- First posted on arXiv 29 Apr 2022.

## Builds on (most cited first, 23 of 23)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [RoBERTa: A Robustly Optimized BERT Pretraining Approach](https://mlsanity.com/paper/1907.11692/) (2019)
- [Intriguing properties of neural networks](https://mlsanity.com/paper/1312.6199/) (2013)
- [DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter](https://mlsanity.com/paper/1910.01108/) (2019)
- [A Broad-Coverage Challenge Corpus for Sentence Understanding through Inference](https://mlsanity.com/paper/1704.05426/) (2017)
- [Language Models as Knowledge Bases?](https://mlsanity.com/paper/1909.01066/) (2019)
- [Right for the Wrong Reasons: Diagnosing Syntactic Heuristics in Natural Language Inference](https://mlsanity.com/paper/1902.01007/) (2019)
- [Is BERT Really Robust? A Strong Baseline for Natural Language Attack on Text Classification and Entailment](https://mlsanity.com/paper/1907.11932/) (2019)
- [ZOO: Zeroth Order Optimization based Black-box Attacks to Deep Neural Networks without Training Substitute Models](https://mlsanity.com/paper/1708.03999/) (2017)
- [BERT-ATTACK: Adversarial Attack Against BERT Using BERT](https://mlsanity.com/paper/2004.09984/) (2020)
- [Black-box Generation of Adversarial Text Sequences to Evade Deep Learning Classifiers](https://mlsanity.com/paper/1801.04354/) (2018)
- [Transformers as Soft Reasoners over Language](https://mlsanity.com/paper/2002.05867/) (2020)
- [Negated and Misprimed Probes for Pretrained Language Models: Birds Can Talk, But Cannot Fly](https://mlsanity.com/paper/1911.03343/) (2019)
- [Gradient-based Adversarial Attacks against Text Transformers](https://mlsanity.com/paper/2104.13733/) (2021)
- [ReClor: A Reading Comprehension Dataset Requiring Logical Reasoning](https://mlsanity.com/paper/2002.04326/) (2020)
- [Neural Variational Inference and Learning in Belief Networks](https://mlsanity.com/paper/1402.0030/) (2014)
- [OpenAttack: An Open-source Textual Adversarial Attack Toolkit](https://mlsanity.com/paper/2009.09191/) (2020)
- [PRover: Proof Generation for Interpretable Reasoning over Rules](https://mlsanity.com/paper/2010.02830/) (2020)
- [Measuring Systematic Generalization in Neural Proof Generation with Transformers](https://mlsanity.com/paper/2009.14786/) (2020)
- [Exploring BERT's Sensitivity to Lexical Cues using Tests from Semantic Priming](https://mlsanity.com/paper/2010.03010/) (2020)
- [Leap-Of-Thought: Teaching Pre-Trained Models to Systematically Reason Over Implicit Knowledge](https://mlsanity.com/paper/2006.06609/) (2020)
- [Generating Natural Language Adversarial Examples on a Large Scale with Generative Models](https://mlsanity.com/paper/2003.10388/) (2020)
- [Generating Semantically Valid Adversarial Questions for TableQA](https://mlsanity.com/paper/2005.12696/) (2020)

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

- [RECKONING: Reasoning through Dynamic Knowledge Encoding](https://mlsanity.com/paper/2305.06349/) (2023)
- [Assessing Logical Reasoning Capabilities of Encoder-Only Transformer Models](https://mlsanity.com/paper/2312.11720/) (2023)

## Abstract (arXiv)

Recent efforts within the AI community have yielded impressive results towards "soft theorem proving" over natural language sentences using language models. We propose a novel, generative adversarial framework for probing and improving these models' reasoning capabilities. Adversarial attacks in this domain suffer from the logical inconsistency problem, whereby perturbations to the input may alter the label. Our Logically consistent AdVersarial Attacker, LAVA, addresses this by combining a structured generative process with a symbolic solver, guaranteeing logical consistency. Our framework successfully generates adversarial attacks and identifies global weaknesses common across multiple target models. Our analyses reveal naive heuristics and vulnerabilities in these models' reasoning capabilities, exposing an incomplete grasp of logical deduction under logic programs. Finally, in addition to effective probing of these models, we show that training on the generated samples improves the target model's performance.