Disi Ji

About

I recently joined Facebook as a research scientist. I am broadly intereted in developing new methods for quantifying and reducing uncertainty of models with techniques from probabilistic learning and statistics. In particular, I am passionate about reliability and fairness of machine learning systems, and their applications to healthcare.

I obtained my Ph.D. degree in computer science from the University of California, Irvine in 2020. Padhraic Smyth was my advisor. Prior to that, I obtained my undergrad degree in mathematics from Fudan University (Shanghai, China) in 2015. I interned at Google Cambridge and Facebook New York before.

[Link to CV]

Research

PhD Thesis

• Label-efficient Bayesian Assessment of Black-box Classifiers. [slides]

Publications

• Bounding the Performance of Human-Machine Classifier Ensembles.
  Gavin Kerrigan, Disi Ji, Padhraic Smyth, Mark Steyvers.
  Presented at Southern California Machine Learning Symposium, 2021
  [In submission]
• Active Bayesian Assessment for Black-Box Classifiers.
  Disi Ji, Robert Logan, Padhraic Smyth, Mark Steyvers.
  35th AAAI Conference on Artificial Intelligence (AAAI), 2021
  [Conference] [Link to arXiv] [slides] [poster] [code] [data]
• Can I Trust My Fairness Metric? Assessing Fairness with Unlabeled Data and Bayesian Inference.
  Disi Ji, Padhraic Smyth, Mark Steyvers.
  34th Conference on Neural Information Processing Systems (NeurIPS), 2020
  [Conference] [Link to arXiv] [slides] [poster] [code] [data]
• Optimization of Automated Gating for Clinical Diagnosis using Discriminative Gates.
  Disi Ji, Preston Putzel, Yu Qian, Richard Scheuermann, Jack D. Bui, Huan-You Wang, and Padhraic Smyth.
  Cytometry: Part A, 2019
  [Journal] [Link to journal] [code]
• Bayesian Evaluation of Black-Box Classifiers.
  Disi Ji, Robert Logan, Padhraic Smyth, Mark Steyvers.
  Uncertainty and Robustness in Deep Learning, ICML 2019.
  [Workshop][Spotlight talk] [pdf] [poster]
• Learning Discriminative Gating Representations for Cytometry Data.
  Disi Ji, Preston Putzel, Yu Qian, Richard Scheuermann, Jack D. Bui, Huan-You Wang, and Padhraic Smyth.
  Workshop on Computational Biology, ICML 2019
  [Workshop] [poster]
• Bayesian Trees for Automated Cytometry Data Analysis.
  Disi Ji, Eric Nalisnick, Yu Qian, Richard Scheuermann, Padhraic Smyth.
  In Proceedings of Machine Learning for Healthcare (MLHC), 2018
  [Conference] [Link to proceedings] [poster] [slides] [code]
• Mondrian Processes for Flow Cytometry Analysis.
  Disi Ji , Eric Nalisnick, and Padhraic Smyth.
  Machine Learning for Health, NIPS 2017
  [Workshop] [pdf]

Medical Abstracts

• Differentiable Gates: Automated Gating for Clinical Diagnosis using Supervised Machine Learning.
  Disi Ji, Preston Putzel, Yu Qian, Richard Scheuermann, Jack D. Bui, Huan-You Wang, and Padhraic Smyth.
  CYTO, Vancouver, Canada, June 22- 26, 2019.
• Interpretable Automated Gating and Classification of Mass Cytometry Data using Machine Learning and Expert Knowledge.
  Disi Ji, Eric Nalisnick, Yu Qian, Richard Scheuermann, Padhraic Smyth.
  CYTO, Prague, Czech, April 28- May 2, 2018.
• Computational Analysis of Cytometry Data Using the FlowGate Cyberinfrastructure.
  Yu Qian, Ivan Chang, Peter Acs, Holden T. Maecker, Michael Reich, Robert Sinkovits, Disi Ji, Padhraic Smyth, Kim Lu, Frank Zaldivar, Dan Cooper, Jill Mesirov, Richard Scheuermann.
  CYTO, Prague, Czech, April 28- May 2, 2018.

Academic Service

• Reviewer: ICML
• PC member, Uncertainty & Robustness in Deep Learning Workshop, ICML 2020
• PC member, Bayesian Deep Learning Workshop, NeurIPS 2019

Teaching

• Teaching assistant: COMPSCI 260 Fundamentals of the Design and Analysis of Algorithms, 2020 Winter
• Teaching assistant: COMPSCI 273A Machine Learning, 2019 Fall
• Instructor: Deep learning with Python, Data Science Initiative Workshop
• Reader: COMPSCI 161 Design and Analysis of Algorithms