PeopleofSuperintelligence

Researcher

Ruixuan Xiao

Ruixuan Xiao is a machine learning researcher whose papers cover partial and noisy labels, active learning with language models, and LLM-based agents. The 2022 paper PiCO, written with Haobo Wang, Yixuan Li, Lei Feng, Gang Niu, Gang Chen and Junbo Zhao, was an ICLR 2022 oral and received an Outstanding Paper Honorable Mention. Xiao went on to first-author ProMix at IJCAI 2023, FreeAL at EMNLP 2023 and the FlowBench agent benchmark in 2024. In 2025, they coauthored the ACL survey OS Agents while affiliated with Zhejiang University.

Timeline

  1. 2022

    PiCO, an ICLR 2022 oral

    In 2022, they coauthor PiCO: Contrastive Label Disambiguation for Partial Label Learning with Haobo Wang, Yixuan Li, Lei Feng, Gang Niu, Gang Chen and Junbo Zhao. The framework pairs a contrastive learning module with a class prototype-based label disambiguation algorithm, and it is presented as an ICLR 2022 oral.

  2. 2022

    An ICLR Honorable Mention

    In 2022, ICLR names PiCO one of its Outstanding Paper Honorable Mentions, and Xiao is the second author. The other honorable mentions go to papers on graph over-squashing and on structured state space models.

  3. 2022–2024

    PiCO+ for noisy labels

    They coauthor PiCO+, the extended version of PiCO, which an arXiv revision of November 2022 presents for noisy partial label learning, where the ground truth may be missing from the candidate set. It selects clean samples by distance and trains robust classifiers with semi-supervised contrastive learning, and the PiCO repository lists it as TPAMI 2024.

  4. 2023

    ProMix at IJCAI

    In 2023, they are first author of ProMix at IJCAI, written with Yiwen Dong, Haobo Wang, Lei Feng, Runze Wu, Gang Chen and Junbo Zhao. The learning-with-noisy-labels method uses a matched high confidence selection technique to expand a clean sample set, and it reports an average improvement of 2.48% on the CIFAR-N dataset.

  5. December 2023

    FreeAL at EMNLP 2023

    In December 2023, they are first author of FreeAL, presented at EMNLP 2023 in Singapore. In this collaborative framework, an LLM serves as an active annotator while a downstream smaller model filters high-quality in-context samples, and experiments on eight benchmark datasets show better zero-shot performance without human supervision.

  6. 2024

    The FlowBench agent benchmark

    In June 2024, they are first author of FlowBench, written with Wentao Ma, Ke Wang, Yuchuan Wu, Junbo Zhao, Haobo Wang, Fei Huang and Yongbin Li. FlowBench is a benchmark for workflow-guided planning of LLM-based agents that covers 51 scenarios from 6 domains.

  7. 2025

    The OS Agents survey

    In 2025, they coauthor OS Agents: A Survey on MLLM-based Agents for Computer, Phone and Browser Use, led by Xueyu Hu and published at ACL 2025. They work on the survey with Zhejiang University as their affiliation.

  1. The company share of AI researchers’ top papers was 31% in 2020–2024 and 8% in 1970–1974Includes Zhejiang University. We estimate where the people shaping AI worked when their most-cited papers appeared: the company share was 31% in 2020–2024 and 8% in 1970–1974, and Google led all organizations in 2020–2024.
  2. The median h-index is 48 at universities and 14 at frontier AI labsIncludes Zhejiang University. The median h-index is 48 at universities and 14 at frontier AI labs. We compare publishing careers, recent citations, dual appointments and job moves across 2,420 university researchers and 308 frontier-lab researchers.
  3. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other universityIncludes Zhejiang University. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other university. We compare doctorate and 5+ year employment links, career timing and current workplaces for 187 major-prize and 1,111 best-paper winners.
  4. More people shaping AI studied at Massachusetts Institute of Technology than at any other institutionIncludes Zhejiang University. We estimate where the people shaping AI studied and where their careers led. Massachusetts Institute of Technology leads with 467 former students, including 291 recorded doctorates.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Ruixuan Xiao yet. Send us a free-licensed photo.

  1. PiCO: Contrastive Label Disambiguation for Partial Label Learning | ICLR 2022 Oral iclr.cc. Checked 22 October 2026.
  2. GitHub - hbzju/PiCO: Contrastive Label Disambiguation for Partial Label Learning (ICLR 2022 Oral) github.com. Checked 22 October 2026.
  3. Announcing the ICLR 2022 Outstanding Paper Award Recipients blog.iclr.cc. Checked 22 October 2026.
  4. PiCO+: Contrastive Label Disambiguation for Robust Partial Label Learning arxiv.org. Checked 22 October 2026.
  5. ProMix: Combating Label Noise via Maximizing Clean Sample Utility | IJCAI ijcai.org. Checked 22 October 2026.
  6. FreeAL: Towards Human-Free Active Learning in the Era of Large Language Models - ACL Anthology aclanthology.org. Checked 22 October 2026.
  7. FlowBench: Revisiting and Benchmarking Workflow-Guided Planning for LLM-based Agents arxiv.org. Checked 22 October 2026.
  8. OS Agents: A Survey on MLLM-based Agents for Computer, Phone and Browser Use - ACL Anthology aclanthology.org. Checked 22 October 2026.
  9. OS Agents: A Survey on MLLM-based Agents for General Computing Devices Use arxiv.org. Checked 22 October 2026.