PeopleofSuperintelligence

Researcher

Jiahua Dong

Jiahua Dong is a computer vision researcher whose AI work spans medical imaging, learning across data shifts, and embodied systems. He pursued doctoral study at the Shenyang Institute of Automation from 2017 to 2024 and is a postdoctoral researcher at MBZUAI. Between 2019 and 2021, he coauthored papers on weakly supervised endoscopic lesion segmentation and multi-source-free domain adaptation, and his 2022 study of out-of-distribution detection earned a NeurIPS Outstanding Paper award. His 2023 work addressed federated incremental segmentation, his 2025 BranchLoRA paper studied continual instruction tuning for multimodal models, and his 2026 papers explored pixel-aware robot action and medical knowledge graphs.

Timeline

  1. 2017–2024

    PhD study at Shenyang

    He pursues doctoral study at the Shenyang Institute of Automation from 2017 to 2024.

  2. 2019

    Weakly supervised lesion segmentation at ICCV

    In 2019, he is first author of the ICCV paper Semantic-Transferable Weakly-Supervised Endoscopic Lesions Segmentation, which transfers knowledge from labeled disease segmentation tasks to weakly supervised endoscopic lesion segmentation.

  3. 2020

    Medical-image domain adaptation

    In 2020, he coauthors Weakly-Supervised Cross-Domain Adaptation for Endoscopic Lesions Segmentation, which studies weakly supervised learning for medical lesion segmentation across domains.

  4. 2021

    Adapting across source domains

    In 2021, he coauthors Confident Anchor-Induced Multi-Source Free Domain Adaptation. The method transfers knowledge from multiple labeled domains to an unlabeled target using pretrained source models without source data.

  5. 2022

    An Outstanding Paper at NeurIPS

    He coauthors Is Out-of-Distribution Detection Learnable? in 2022. The paper studies PAC learning theory for OOD detection and receives a NeurIPS Outstanding Paper award.

  6. 2023

    Federated incremental segmentation

    He and five coauthors propose Forgetting-Balanced Learning in a 2023 CVPR paper on federated incremental semantic segmentation. The model addresses forgetting within local clients and across clients as they add new classes under privacy protection.

  7. 2025

    BranchLoRA for instruction tuning

    In 2025, he coauthors BranchLoRA, a framework for continual instruction tuning of multimodal large language models. Its flexible tuning-freezing mechanism lets branches specialize in intra-task knowledge and collaborate across tasks.

  8. Since 2024

    Postdoc at MBZUAI

    He takes up a postdoctoral role at MBZUAI in 2024. In 2026, he coauthors PixelVLA, a vision-language-action model that combines pixel-level reasoning with text and visual prompts for robotic manipulation. In August 2026, he coauthors MedKGent, an agent framework that builds a temporal medical knowledge graph from PubMed abstracts.

Career

  1. Since 2024Mohamed bin Zayed University of Artificial IntelligencePostdoctoral researcher

Education

  1. 2024Chinese Academy of SciencesDoctoral study

From the biography.

  1. The United States employs an estimated 48% of people shaping AIIncludes Mohamed bin Zayed University of Artificial Intelligence. We estimate where the people shaping AI work today, how universities, companies and research institutes split across countries, and how country shares have changed since the 1960s.
  2. The company share of AI researchers’ top papers was 31% in 2020–2024 and 8% in 1970–1974Includes Mohamed bin Zayed University of Artificial Intelligence. 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.
  3. Google has the highest combined h-index among AI organizationsIncludes Mohamed bin Zayed University of Artificial Intelligence. We estimate that Google holds the highest combined h-index at 8,016, while universities account for 73% of the deduplicated total across AI organizations.
  4. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other universityIncludes Mohamed bin Zayed University of Artificial Intelligence. 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.

Colleagues

People with dated positions at the same organization in the same years.

Sources and credits

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

  1. Jiahua Dong (Mohamed bin Zayed University of Artificial Intelligence) lacuna.tiptreesystems.com. Checked 22 October 2026.
  2. ICCV 2019 Open Access Repository openaccess.thecvf.com. Checked 22 October 2026.
  3. [2012.04170] Weakly-Supervised Cross-Domain Adaptation for Endoscopic Lesions Segmentation arxiv.org. Checked 22 October 2026.
  4. Confident Anchor-Induced Multi-Source Free Domain Adaptation papers.neurips.cc. Checked 22 October 2026.
  5. NeurIPS 2022 Awards neurips.cc. Checked 22 October 2026.
  6. CVPR 2023 Open Access Repository openaccess.thecvf.com. Checked 22 October 2026.
  7. Enhancing Multimodal Continual Instruction Tuning with BranchLoRA - ACL Anthology aclanthology.org. Checked 22 October 2026.
  8. PixelVLA — Project Page wenqiliang.github.io. Checked 22 October 2026.
  9. MedKGent: a large language model agent framework for constructing temporally evolving medical knowledge graph | npj Digital Medicine nature.com. Checked 22 October 2026.