English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Reasoning has been a central topic in artificial intelligence from thebeginning. The recent progress made on distributed representation and neuralnetworks continues to improve the state-of-the-art performance of naturallanguage inference. However, it remains an open question whether the modelsperform real reasoning to reach their conclusions or rely on spuriouscorrelations. Adversarial attacks have proven to be an important tool to helpevaluate the Achilles' heel of the victim models. In this study, we explore thefundamental problem of developing attack models based on logic formalism. Wepropose NatLogAttack to perform systematic attacks centring around naturallogic, a classical logic formalism that is traceable back to Aristotle'ssyllogism and has been closely developed for natural language inference. Theproposed framework renders both label-preserving and label-flipping attacks. Weshow that compared to the existing attack models, NatLogAttack generates betteradversarial examples with fewer visits to the victim models. The victim modelsare found to be more vulnerable under the label-flipping setting. NatLogAttackprovides a tool to probe the existing and future NLI models' capacity from akey viewpoint and we hope more logic-based attacks will be further explored forunderstanding the desired property of reasoning.

NatLogAttack: A Framework for Attacking Natural Language Inference Models with Natural Logic