/Paper-Review

This is for the papers I review every week

Paper-Review

This repo contains notes and short summaries of some DNN security and Software Engineering & AI related papers I come across.

Attacks

Backdoor Attack

Watermarking

  • Leveraging Unlabeled Data for Watermark Removal of Deep Neural Networks (ICML 2019): [Paper] [Notes]
  • Turning Your Weakness Into a Strength: Watermarking Deep Neural Networks by Backdooring (usenix 2018): [Paper] [Code] [Blog] [Notes] [Citation]

Textual Adversarial Attack

Char-level Attack

  • TEXTBUGGER: Generating Adversarial Text Against Real-world Applications (NDSS 2019): [Paper]
  • Text processing like humans do: Visually attacking and shielding NLP systems (NAACL 2019): [Paper]
  • Black-box generation of adversarial text sequences to evade deep learning classifiers (S&P 2018): [Paper]
  • Hotflip: White-box adversarial examples for text classification (ACL 2018): [Paper]

Word-level Attack

  • Generating natural language adversarial examples (EMNLP 2018): [Paper]
  • Deep text classification can be fooled (IJCAI 2018): [Paper]
  • Generating Fluent Adversarial Examples for Natural Languages (ACL 2019): [Paper]
  • Generating natural language adversarial examples through probability weighted word saliency (ACL 2019): [Paper]

Backdoor Detection and Mitigation

  • Neural Cleanse: Identifying and Mitigating Backdoor Attacks in Neural Networks (S&P 2019): [Paper] [Code] [Notes] [Citation]

  • STRIP: A Defence Against Trojan Attacks on Deep Neural Networks (2019 ACSAC): [Paper] [Code] [Notes] [Citation]

  • Design and Evaluation of a Multi-Domain Trojan Detection Method on Deep Neural Networks (2019): [Paper] [Notes] [Citation]

  • DeepInspect: A Black-box Trojan Detection and Mitigation Framework for Deep Neural Networks (IJCAI 2019): [Paper] [Notes] [Citation]

  • ABS: Scanning Neural Networks for Back-doors by Artificial Brain Stimulation (CCS 2019): [Paper] [Citation]

  • TABOR: A Highly Accurate Approach to Inspecting and Restoring Trojan Backdoors in AI Systems (2019): [Paper] [Notes] [Citation]

  • Detecting Backdoor Attacks on Deep Neural Networks by Activation Clustering (2019 AAAI): [Paper] [Code] [Notes] [Citation]

Software Engineering and AI

DNN Testing

  • DEEPSEC: deciding equivalence properties in security protocols theory and practice (S&P 2018): [Paper]
  • Guiding deep learning system testing using surprise adequacy (ICSE 2019): [Paper]
  • DeepGauge: Multi-granularity testing criteria for deep learning systems (ASE 2018): [Paper]
  • DeepXplore: Automated whitebox testing of deep learning systems (SOSP 2017): [Paper]
  • DeepTest: Automated testing of deep-neural-network-driven autonomous cars (ICSE 2018): [Paper]
  • DeepHunter: a coverage-guided fuzz testing framework for deep neural networks (ISSTA 2019): [Paper]
  • There is limited correlation between coverage and robustness for deep neural networks (ASE 2019): [Paper]

NMT Testing

  • Structure-Invariant Testing for Machine Translation (ICSE 2020): [Paper]
  • Automatic Testing and Improvement of Machine Translation (ICSE 2020): [Paper]
  • Metamorphic Testing for Machine Translations: MT4MT (ASWEC 2018): [Paper]