AI Paper Atlas

Research map / Adversarial robustness and security / Certified and empirical robustness

Learning to Repair Software Vulnerabilities with Generative Adversarial Networks

Jacob Harer, Onur Ozdemir, Tomo Lazovich, Christopher Reale, Rebecca Russell, Louis Kim, peter chin

NeurIPS 2018 · Adversarial robustness and security › Certified and empirical robustness · approach: Adversarial training with feature consistency · 43 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add Auto or Freq self-regularization to the unregularized GAN (ablation)Proposed GAN · SortingSeq. Acc.: 82.8 -> 98.9; 82.8 -> 99.3Improved
add curriculum learning without self-regularization (ablation)Proposed GAN · SortingSeq. Acc.: 82.8 -> 81.5Hurt
add Auto regularization without curriculum (ablation)Proposed GAN · CFGCFG Acc: 98.0 -> 96.5Hurt
add Freq regularization without curriculum (ablation)Proposed GAN · CFGCFG Acc: 98.0 -> 97.5Hurt
add curriculum learning without self-regularization (ablation)Proposed GAN · SATE IV pairedBLEU-4: 84.2 -> 88.3Improved
add Auto regularization without curriculum (ablation)Proposed GAN · SATE IV pairedBLEU-4: 84.2 -> 85.7Improved
add Freq regularization without curriculum (ablation)Proposed GAN · SATE IV pairedBLEU-4: 84.2 -> 86.2Improved
add curriculum learning without self-regularization (ablation)Proposed GAN · SATE IV unpairedBLEU-4: 79.3 -> 81.1Improved
add Auto regularization without curriculum (ablation)Proposed GAN · SATE IV unpairedBLEU-4: 79.3 -> 79.2No clear effect
use unpaired rather than paired training dataProposed GAN Base · SATE IVBLEU-4: 84.2 -> 79.3Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy