Lunjia Hu
I am a final-year Ph.D. candidate in the Computer Science Department at Stanford University. I am very fortunate to be co-advised by Moses Charikar and Omer Reingold. Before Stanford, I was an undergraduate in the Yao Class at Tsinghua University, where I received B.Eng. in computer science and B.S. in mathematics.
I work on advancing the theoretical foundations of trustworthy machine learning, addressing fundamental questions about interpretability, fairness, robustness, and uncertainty quantification.
Preprints:
Predict to Minimize Swap Regret for All Payoff-Bounded Tasks. [arXiv]
Lunjia Hu, Yifan Wu
Testing Calibration in Subquadratic Time. [arXiv]
Lunjia Hu, Kevin Tian, Chutong Yang
On Computationally Efficient Multi-Class Calibration. [arXiv]
Parikshit Gopalan, Lunjia Hu, Guy N. Rothblum
Conference papers:
Jarosław Błasiok, Parikshit Gopalan, Lunjia Hu, Adam Tauman Kalai, Preetum Nakkiran
ITCS 2024
When Does Optimizing a Proper Loss Yield Calibration? [arXiv]
Jarosław Błasiok, Parikshit Gopalan, Lunjia Hu, Preetum Nakkiran
NeurIPS 2023 (Spotlight)
Simple, Scalable and Effective Clustering via One-Dimensional Projections. [arXiv]
Moses Charikar, Monika Henzinger, Lunjia Hu, Maximilian Vötsch, Erik Waingarten
NeurIPS 2023
Generative Models of Huge Objects. [arXiv]
Lunjia Hu, Inbal Rachel Livni Navon, Omer Reingold
CCC 2023
Omnipredictors for Constrained Optimization. [arXiv] [talk at the Simons Institute]
Lunjia Hu, Inbal Rachel Livni Navon, Omer Reingold, Chutong Yang
ICML 2023
Jarosław Błasiok, Parikshit Gopalan, Lunjia Hu, Preetum Nakkiran
STOC 2023
Comparative Learning: A Sample Complexity Theory for Two Hypothesis Classes. [arXiv] [video] [Charlotte's talk at the Simons Institute]
Lunjia Hu, Charlotte Peale
ITCS 2023
Loss Minimization through the Lens of Outcome Indistinguishability. [arXiv] [video by Michael]
Parikshit Gopalan, Lunjia Hu, Michael P. Kim, Omer Reingold, Udi Wieder
ITCS 2023
Subspace Recovery from Heterogeneous Data with Non-isotropic Noise. [arXiv]
John Duchi, Vitaly Feldman, Lunjia Hu, Kunal Talwar
NeurIPS 2022
Metric Entropy Duality and the Sample Complexity of Outcome Indistinguishability. [arXiv]
Lunjia Hu, Charlotte Peale, Omer Reingold
ALT 2022
E. M. Gold Best Student Paper Award
An Improved Local Search Algorithm for k-Median. [arXiv]
Vincent Cohen-Addad, Anupam Gupta, Lunjia Hu, Hoon Oh, David Saulpic
SODA 2022
Near-Optimal Explainable k-Means for All Dimensions. [arXiv] [talk at the IDEAL Workshop]
Moses Charikar, Lunjia Hu
SODA 2022
Robust Mean Estimation on Highly Incomplete Data with Arbitrary Outliers. [arXiv]
Lunjia Hu, Omer Reingold
AISTATS 2021
Approximation Algorithms for Orthogonal Non-negative Matrix Factorization. [arXiv]
Moses Charikar, Lunjia Hu
AISTATS 2021
The Power of Many Samples in Query Complexity. [conference version] [arXiv]
Andrew Bassilakis, Andrew Drucker, Mika Göös, Lunjia Hu, Weiyun Ma, Li-Yang Tan
ICALP 2020
Towards Understanding Learning Representations: To What Extent Do Different Neural Networks Learn the Same Representation. [conference version] [arXiv] [spotlight talk]
Liwei Wang, Lunjia Hu, Jiayuan Gu, Zhiqiang Hu, Yue Wu, Kun He, John Hopcroft
NeurIPS 2018 (Spotlight)
Active Tolerant Testing. [conference version] [arXiv] [conference talk]
Avrim Blum, Lunjia Hu.
COLT 2018
Quadratic Upper Bound for Recursive Teaching Dimension of Finite VC Classes. [conference version] [arXiv]
Lunjia Hu, Ruihan Wu, Tianhong Li, Liwei Wang.
COLT 2017
Capacitated Center Problems with Two-Sided Bounds and Outliers. [conference version] [arXiv]
Hu Ding, Lunjia Hu, Lingxiao Huang, Jian Li.
WADS 2017
Professional Activities
Conference & Journal Reviewing: ALT 2019, NeurIPS (2019, 2020, 2023), SODA (2022, 2023), STOC (2022, 2024), ESA 2022, FORC 2023, FOCS 2023, ITCS 2024, COLT 2024, JMLR
TAing:
The Practice of Theory Research (CS163), Winter 2020-2021, Stanford.
I was selected as one of the top 5% course assistants in the Computer Science Department at Stanford University in Winter 2020-2021.Randomized Algorithms and Probabilistic Analysis (CS265/CME309), Fall 2020-2021, Stanford.
Other:
I wrote two blog posts on Stanford’s CS theory research blog (Theory Dish) about the sunflower lemma and the backpropagation algorithm.
Contact Information
Email: [first name] [at] stanford [dot] edu