Aaron Chan

PhD Candidate at USC   •   aarzchan :slightly_smiling_face: gmail :upside_down_face: com


prof_pic.jpg

Hi! :wave: I’m a computer science PhD candidate in the INK Lab at the University of Southern California (USC), where I’m advised by Prof. Xiang Ren. I’m also a member of USC’s Information Sciences Institute (ISI).

My research is at the intersection of machine learning and natural language processing. In particular, I’m excited about:

  • model explainability: explaining language model behavior more faithfully, plausibly, and efficiently.
  • explanation-based learning: operationalizing explanations to improve language model generalization and decision-making.

In the past, I’ve worked as a research intern at Meta AI and an engineering intern at Google.

Before coming to USC, I earned a master’s degree in robotics from the University of Pennsylvania and a bachelor’s degree in electrical engineering from the University of Maryland, College Park.

On a personal note, I grew up in the suburbs of Washington, DC and now live with my wife in Los Angeles, CA. Outside of work, you might find me playing/watching basketball (go Wizards!), skiing, hiking, board gaming, or foodie-ing (:sushi::curry::crab::pizza:). Plus, I’m a huge fan of the TV shows Seinfeld and BoJack Horseman.


News

May 15, 2022 UNIREX was accepted to ICML 2022! :confetti_ball:
May 10, 2022 After completing my PhD (expected Fall 2022), I’ll be joining Meta AI as a research scientist on the AI Integrity Team! :man_technologist:
May 09, 2022 ER-Test was accepted to the TrustNLP Workshop @ NAACL 2022! :tada:
Apr 30, 2022 I’m presenting UNIREX today at the Socially Responsible ML Workshop @ ICLR 2022. Come find me at Poster D2!
Mar 31, 2022 UNIREX was accepted to the BigScience Workshop @ ACL 2022! :tada:
Mar 27, 2022 UNIREX was accepted to the Socially Responsible ML Workshop @ ICLR 2022! :tada:
Mar 11, 2022 I passed my PhD thesis proposal! :ballot_box_with_check: Check out my slides here.
Jan 10, 2022 In Spring 2022, I’m a TA for CSCI 566 (Deep Learning and its Applications) at USC.
Dec 16, 2021 A preprint of my Meta AI intern project, UNIREX, is now up on arXiv.
Dec 09, 2021 I’m presenting SalKG today at NeurIPS 2021. Come chat with me at Poster Session 7!
Sep 28, 2021 SalKG was accepted to NeurIPS 2021! :confetti_ball:
Sep 20, 2021 I started my research internship at Facebook AI Meta AI, where I’ll be working on language model explainability with the AI Integrity Team.
Jul 31, 2021 This summer, I had an awesome time mentoring Wyatt Lake, as part of USC’s SHINE program. Check out Wyatt’s project on explanation-based regularization of BERT models.
Jun 08, 2021 I passed my PhD qualifying exam and am now a PhD candidate! :ballot_box_with_check:

Selected Publications

  1. Preprint
    FRAME: Evaluating Faithfulness Metrics for Free-Text Rationales
    A. Chan, S. Nie, L. Tan, X. Peng, L. Mathias, H. Firooz, M. Sanjabi, and X. Ren
    Preprint, 2022
    PDF
  2. TrustNLP
    ER-Test: Evaluating Explanation Regularization Methods for NLP Models
    B. Joshi*, A. Chan*, Z. Liu, S. Nie, M. Sanjabi, H. Firooz, and X. Ren
    TrustNLP Workshop at NAACL, 2022
    arXiv PDF
  3. ICML
    UNIREX: A Unified Learning Framework for Language Model Rationale Extraction
    A. Chan, M. Sanjabi, L. Mathias, L. Tan, S. Nie, X. Peng, X. Ren, and H. Firooz
    International Conference on Machine Learning (ICML), 2022
    arXiv PDF
  4. NeurIPS
    SalKG: Learning From Knowledge Graph Explanations for Commonsense Reasoning
    A. Chan, J. Xu, B. Long, S. Sanyal, T. Gupta, and X. Ren
    Advances in Neural Information Processing Systems (NeurIPS), 2021
    arXiv PDF Code
  5. ACL Findings
    Learning Contextualized Knowledge Structures for Commonsense Reasoning
    J. Yan, M. Raman, A. Chan, T. Zhang, R. Rossi, H. Zhao, S. Kim, N. Lipka, and X. Ren
    Findings of the Annual Meeting of the Association for Computational Linguistics (ACL), 2021
    arXiv PDF Code
  6. ICLR
    Learning to Deceive Knowledge Graph Augmented Models via Targeted Perturbation
    M. Raman, A. Chan*, S. Agarwal*, P. Wang, H. Wang, S. Kim, R. Rossi, H. Zhao, N. Lipka, and X. Ren
    International Conference on Learning Representations (ICLR), 2021
    arXiv PDF Code