Robustness

We might hope that when faced with unexpected inputs, well-designed software systems would fire off warnings. However, ML systems, which depend strongly on properties of their inputs (e.g. the i.i.d. assumption), tend to fail silently. Faced with distribution shift, we wish (i) to detect and (ii) to quantify the shift, and (iii) to correct our classifiers on the fly—when possible. Unfortunately, absent assumptions, the problem is impossible. Moreover, many modern methods claim robustness but don't make clear which assumptions (if any) are coherently addressed. Our work focuses both on researching foundational principles for building systems that can be relied upon in the wild and developing a body of empirical work to guide practitioners.

• Just Label the Repeats for In-The-Wild Audio-to-Score Alignment

  Irmak Bukey, Michael Feffer, Chris Donahue
  International Society for Music Information Retrieval Conference (ISMIR)

• Red-Teaming for Generative AI: Silver Bullet or Security Theater?

  Michael Feffer, Anusha Sinha, Wesley H. Deng, Zachary Lipton, Hoda Heidari
  AAAI/ACM Conference on Artificial Intelligence, Ethics and Society (AIES)

• Prompt Exploration with Prompt Regression

  Michael Feffer*, Ronald Xu*, Yuekai Sun, Mikhail Yurochkin
  Conference on Language Modeling (COLM)

• GenAudit: Fixing Factual Errors in Language Model Outputs with Evidence

  Kundan Krishna, Sanjana Ramprasad, Prakhar Gupta, Byron Wallace, Zachary Lipton, Jeffrey Bigham
  arXiv (preprint)

• Domain Adaptation under Missingness Shift

  Helen Zhou, Sivaraman Balakrishnan, Zachary Lipton
  Artificial Intelligence and Statistics (AISTATS)

• Deconstructing distributions: A pointwise framework of learning

  Gal Kaplun, Nkhil Ghosh, Saurabh Garg, Boaz Barak, Preetum Nakkiran
  International Conference on Learning Representations (ICLR)

• Chils: Zero-shot image classification with hierarchical label sets

  Zachary Novack, Julian McAuley, Zachary Lipton, Saurabh Garg
  arXiv preprint arXiv:2302.02551

• RLSbench: Domain Adaptation Under Relaxed Label Shift

  Saurabh Garg, Nick Erickson, James Sharpnack, Alex Smola, Sivaraman Balakrishnan, Zachary Lipton
  arXiv preprint arXiv:2302.03020

• Unsupervised Learning under Latent Label Shift

  Manley Roberts*, Pranav Mani*, Saurabh Garg, Zachary Lipton
  Advances in Neural Information Processing Systems (NeurIPS)

• Domain Adaptation under Open Set Label Shift

  Saurabh Garg, Sivaraman Balakrishnan, Zachary Lipton
  Advances in Neural Information Processing Systems (NeurIPS)

• Practical Benefits of Feature Feedback Under Distribution Shift

  Anurag Katakkar*, Clay Yoo*, Weiqin Wang*, Zachary Lipton, Divyansh Kaushik
  BlackBox NLP workshop

• Leveraging Unlabeled Data to Predict Out-of-Distribution Performance

  Saurabh Garg, Sivaraman Balakrishnan, Zachary Lipton, Hanie Sedghi, Behnam Neyshabur
  International Conference on Learning Representations (ICLR)

• Mixture Proportion Estimation and PU Learning: A Modern Approach

  Saurabh Garg, Yifan Wu, Alex Smola, Sivaraman Balakrishnan, Zachary Lipton
  Advances in Neural Information Processing Systems (NeurIPS)

• Off-Policy Risk Assessment in Contextual Bandits

  Audrey Huang, Liu Leqi, Zachary Lipton, Kamyar Azizzadenesheli
  Advances in Neural Information Processing Systems (NeurIPS)

• On the Efficacy of Adversarial Data Collection for Question Answering: Results from a Large-Scale Randomized Study

  Divyansh Kaushik, Douwe Kiela, Zachary Lipton, Wen-tau Yih
  Association for Computational Linguistics and International Joint Conference on Natural Language Processing (ACL-IJCNLP)

• RATT: Leveraging Unlabeled Data to Guarantee Generalization

  Saurabh Garg, Sivaraman Balakrishnan, J. Zico Kolter, Zachary Lipton
  International Conference on Machine Learning (ICML)

• On Proximal Policy Optimization’s Heavy-Tailed Gradients

  Saurabh Garg, Emilio Parisotto, Adarsh Prasad, J. Zico Kolter, Zachary Lipton, Sivaraman Balakrishnan, Ruslan Salakhutdinov, Pradeep Ravikumar
  International Conference on Machine Learning (ICML)

• Explaining The Efficacy of Counterfactually-Augmented Data

  Divyansh Kaushik, Zachary Lipton
  International Conference on Learning Representations (ICLR)

• A Unified View of Label Shift Estimation

  Saurabh Garg, Yifan Wu, Sivaraman Balakrishnan, Zachary Lipton
  Advances in Neural Information Processing Systems (NeurIPS)

• Learning to Deceive with Attention-Based Explanations

  Danish Pruthi, Mansi Gupta, Bhuwan Dhingra, Graham Neubig, Zachary Lipton
  Association for Computational Linguistics (ACL)

• Learning the Difference that Makes a Difference with Counterfactually-Augmented Data

  Divyansh Kaushik, Zachary Lipton
  International Conference on Learning Representations (ICLR)

• Are Perceptually-Aligned Gradients a General Property of Robust Classifiers?

  Simran Kaur, Jeremy Cohen, Zachary Lipton
  Neurips 2019 Science Meets Engineering of Deep Learning Workshop

• Learning Robust Global Representations by Penalizing Local Predictive Power

  Haohan Wang, Songwei Ge, Eric Xing, Zachary Lipton
  Advances in Neural Information Processing Systems (NeurIPS)

• Failing Loudly: An Empirical Study of Methods for Detecting Dataset Shift

  Stefan Rabanser, Stephan Günnemann, Zachary Lipton
  Advances in Neural Information Processing Systems (NeurIPS)

• Domain Adaptation with Asymmetrically-Relaxed Distribution Alignment

  Yifan Wu, Divyansh Kaushik, Ezra Winston, Zachary Lipton
  International Conference on Machine Learning (ICML)

• What is the Effect of Importance Weighting in Deep Learning?

  Jonathon Byrd, Zachary Lipton
  International Conference on Machine Learning (ICML)

• Reinforcement Learning with Unknown Reward Functions

  Benjamin Eysenbach*, Jacob Tyo*, Shane Gu, Ruslan Salakhutdinov, Zachary Lipton, Sergey Levine
  Task-Agnostic Reinforcement Learning Workshop, ICLR

• Learning Robust Representations by Projecting Superficial Statistics Out

  Haohan Wang, Zexue Hu, Zachary Lipton, Eric Xing
  International Conference on Learning Representations (ICLR)

• Learning Noise-Invariant Representations for Robust Speech Recognition

  Davis Liang, Zhiheng Huang, Zachary Lipton
  IEEE Speech and Language Technology (SLT)

• Detecting and Correcting for Label Shift with Black Box Predictors

  Zachary Lipton*, Yu-Xiang Wang*, Alex Smola
  International Conference on Machine Learning (ICML)

• Stochastic Activation Pruning for Robust Adversarial Defense

  Guneet Dhillon Singh, Kamyar Azizzadenesheli, Jeremy Bernstein, Aran Khanna, Zachary Lipton, Anima Anandkumar
  International Conference on Learning Representations (ICLR)