#ICLR2022 How can learning invariant representations help cross-lingual transfer? Please come to our poster session at 12:30pm for more details, Session 5 Room 2. Joint work with ruicheng @elgreco_winter Paper: openreview.net/pdf?id=k7-s5HS… Code: github.com/rxian/domain-a…

Excited that 2/2 of my 1st-author papers get accepted by ICML! Thank my amazing advisors Han Zhao Bo Li ! Provable Domain Generalization via Invariant-Feature Subspace Recovery: arxiv.org/abs/2201.12919 Understanding Gradual Domain Adaptation: arxiv.org/abs/2204.08200

Happy to share our recent work on understanding linear scalarization vs. multi-objective optimization for multitask learning, to appear at #NeurIPS2023: arxiv.org/pdf/2308.13985…. TL;DR: Is linear scalarization always sufficient for MTL? If not, when will it fail? 1/n

Interested in domain adaptation in neural IR? Check out this amazing #NeurIPS2023 paper (selected for spotlight presentation) on Learning List-Level Domain-Invariant Representations for Ranking. arxiv.org/pdf/2212.10764… Great work by ruicheng, Honglei Zhuang et al.

🎉 Thrilled to announce our work “Robust Multi-Task Learning with Excess Risks”has been accepted at #ICML2024! We introduce ExcessMTL, an excess risk based adaptive task weighting method that is robust to label noise. 1/n

How to ensure fairness (statistical parity) and privacy (DP) simultaneously? What are the costs of privacy and fairness upon accuracy? Excited to share our #ICML2024 work answering the two questions above! paper: arxiv.org/pdf/2405.04034 code: github.com/rxian/fair-reg…

Excited to share that I am joining the University of Washington ECE as an Assistant Professor in September. We are recruiting postdoc, PhD students (2025 cycle), as well as master and undergraduate researchers! Check out our research at yyueluo.com.

Proud advisor moment! Huge congrats to Dr. Ruicheng Xian ruicheng on a truly wonderful thesis, and many thanks to Gautam Kamath, Aaron Roth, Arindam, and Hanghang for serving on his PhD committee. Ruicheng’s PhD work characterizes the fundamental trade-off between group

GPT-5.5, not fully saturating the TikZ unicorn test yet but getting awfully close ... (yes this is actual TikZ code, I personally find it so unbelievable that I'm putting the code below for anyone to verify for themself)