Jianhan Wu

Jianhan Wu

Algorithm Engineer

Jianhan Wu has graduated from the University of Science and Technology of China (USTC) with a master degree. He currently works as an Algorithm Engineer, and has long-term research and industrial-implementation experience in LMs, embodied intelligence, privacy-preserving computing. His research interests cover lightweight adaptation of large models, multimodal perception and understanding, optimization of embodied-intelligence systems, as well as edge-cloud collaborative inference. In terms of research projects, he has served as a key core member participating in multiple provincial-municipal key research tasks. For the Key R&D Program of Guangdong Provincial Department of Science and Technology, he tackles key technologies including on-device model deployment and edge-cloud collaborative intelligent computing. For the Shenzhen-Hong Kong Joint Funding Project (2nd principal contributor), he conducts research and development on privacy-preserving federated large-model systems. He has filed more than 50 invention patents, among which 19 have been granted. These patents cover core technical areas such as lightweight compression for large models, efficient processing of multimodal perception data, edge-cloud inference acceleration, and perception-reasoning optimization for embodied intelligence. Multiple patented technologies have been verified through industrial deployment in real-world business scenarios including smart finance and intelligent interaction. From the academic perspective, he has published over 10 papers at well-known international conferences including ICCV, ICASSP, INTERSPEECH, ACML and IJCNN.

Interests
  • Embodied AI
  • Large Models
  • Federated Learning
  • Deep Learning
Awards
  • 4. 2026 XinZhi Award - Excellent Case Selection for Financial Institution Digital Intelligence Transformation TOP10 Excellent Award (2nd Contributor) / 2026 鑫智奖-金融机构数智化转型优秀案例评选-TOP10优秀奖
  • 3. 2025 ‘Data Element X’ Competition National Finals Second Prize (Contributor) / 2025 年“数据要素X”大赛全国总决赛二等奖《基于数据要素驱动的保险风控服务项目》
  • 2. 2025 2nd CCF China Digital Finance Conference ‘Digital Finance Exploration and Innovation Excellent Case’ / 2025第二届CCF中国数字金融大会“数字金融探索创新优秀案例”
  • 1. NVIDIA 10th Hackathon Excellence Award / 英伟达第十届黑客松比赛优秀奖

Publications

  1. FASA: Feedback-Aware Sampling Adaptation for Efficient Diffusion-Based VLA Model, (2026), ✉Corresponding Author, In NAS2026 (CCF-C)
  2. Mita: A Hierarchical Multi-Agent Collaboration Framework with Memory-Integrated and Task Allocation (2026), In ICASSP2026 (CCF-B)
  3. Federated Domain Generalization with Domain-specific Soft Prompts Generation, (2025), †First Author, In ICCV2025 (CCF-A)
  4. Augmentation-induced Consistency Regularization for Classification, (2022), †First Author, In IJCNN2022 (CCF-C)
  5. Improving Human Image Synthesis with Residual Fast Fourier Transformation and Wasserstein Distance, (2022), †First Author, In IJCNN2022 (CCF-C)
  6. Pose Guided Human Image Synthesis with Partially Decoupled GAN, (2022), †First Author, In ACML2022 (CCF-C)
  7. Variational Information Bottleneck for Effective Low-Resource Audio Classification (2021), In INTERSPEECH2021 (CCF-B)

中文期刊文章

  1. 时间锁谜题综述 (2026), 《网络与信息安全学报》(CCF-T1)
  2. 结合少样本逻辑推理的多模态机器人故障解释一致性方法, (2025), ✉Corresponding Author, 《大数据》(CCF-T2)
  3. 联邦学习攻击与防御综述, (2022), †First Author, 《大数据》,8 (05),(CCF-T2)