徐盈辉
复旦大学人工智能创新与产业研究院研究员、博士生导师
手机：
微信：19821197419（助理）
邮箱：
驻地：
徐盈辉
复旦大学人工智能创新与产业研究院研究员、博士生导师
个人履历：
徐盈辉：复旦大学人工智能创新与产业研究院研究员、博士生导师。曾任：阿里巴巴达摩院视觉实验室负责人、研究员，菜鸟网络科技人工智能部负责人、研究员，阿里巴巴搜索事业部基础算法负责人、研究员。徐盈辉的研究方向包括：(1)生成式AI 与逻辑推理的融合：这个研究方向主要探索如何将生成式AI（例如，基于深度学习的模型）和逻辑推理（符号推理技术）相结合，以提高模型的信息准确性和进行高层次的推断和预测。(2)生命科学领域的知识融合：这个方向强调在AI 模型中融入生命科学领域的知识，从而增强模型对生命科学领域问题的理解和解决能力。(3)微观/宏观一致性生命科学大模型研究：这个研究方向主要探索如何构建一种生命科学大模型，这个模型能够理解微观（如单个生物过程）与宏观（如整个生态系统或生物体）的关联和因果关系。(4)第一性原理计算与AI 反馈闭环系统：这个方向研究如何将基于物理规则（第一性原理）的计算和AI 模型结合，形成一个反馈闭环，以便从整体上理解和模拟生命系统。 (5)应用研究：这个方向主要关注如何利用上述技术来解决生命科学中的实际问题，如创新药物的发现、多组学分析、以及个性化疾病诊断等。已发表SCI 论文50余篇；授权发明专利十余项。代表性成果有：(1)带领团队最早引入基于强化学习的排序策略决策模型，相关成果在KDD2018 发表，引起学术界和工业界的广泛关注；(2)带领团队将迁移学习，主动学习，半监督学习等技术合理应用，研发一套有业界领先优势的遥感卫星解译产品；(3)带领团队研发了一套自适应大规模邻域搜索和深度强化学习为基础的分布式启发式优化引擎，在VRP 算法领域权威的评测平台上取得了多项最好成绩，并入围运筹学界的“奥斯卡”2021 Franz Edeleman 杰出成就奖；(4)最早引入的GAN 生成算法到信息检索领域，并在2017年信息检索领域顶会SIGIR 获得最佳论文提名。
Xu Yinghui: Researcher and PhD supervisor at the Institute of Artificial Intelligence Innovation and Industry, Fudan University. Former positions include: Head of the Visual Lab and Researcher at Alibaba DAMO Academy, Head of the AI Department and Researcher at Cainiao Network Technology, Head of Fundamental Algorithms and Researcher at Alibaba Search Business Unit. Xu Yinghui's research areas include: (1) Integration of Generative AI and Logical Reasoning: This research focuses on exploring how to combine generative AI (e.g., deep learning-based models) with logical reasoning (symbolic reasoning techniques) to improve the accuracy of information and enable higher-level inference and prediction. (2) Knowledge Integration in Life Sciences: This direction emphasizes incorporating knowledge from the life sciences into AI models to enhance their understanding and problem-solving capabilities in this field. (3) Research on Micro/Macro Consistent Large-scale Life Science Models: This area explores how to build a life science large model capable of understanding the relationships and causality between micro-level (such as individual biological processes) and macro-level (such as entire ecosystems or organisms) phenomena. (4) First-Principles Computation and AI Feedback Loop Systems: This direction studies how to combine physics-based (first-principles) computation with AI models to form a feedback loop, aiming for a comprehensive understanding and simulation of life systems. (5) Applied Research: This area focuses on using these technologies to address practical problems in life sciences, such as innovative drug discovery, multi-omics analysis, and personalized disease diagnosis. Xu Yinghui has published over 50 SCI papers and holds more than ten authorized invention patents. Representative achievements include: (1) Leading the team to be among the first to introduce a reinforcement learning-based ranking strategy decision model, with related results published at KDD 2018, attracting widespread attention in academia and industry; (2) Leading the team to appropriately apply techniques like transfer learning, active learning, and semi-supervised learning, developing a remote sensing satellite interpretation product with industry-leading advantages; (3) Leading the team to develop a distributed heuristic optimization engine based on adaptive large-scale neighborhood search and deep reinforcement learning, achieving multiple best results on authoritative VRP algorithm uation platforms and shortlisted for the 2021 Franz Edelman Award for Outstanding Achievement in Operations Research; (4) Among the first to introduce GAN-based generation algorithms into the field of information retri, receiving a Best Paper nomination at SIGIR 2017, a top conference in information retri.
邀请老师演讲、授课请致电：19821197419 阎老师[微信同号]
免责声明：以上内容（包括文字、图片、视频）为用户上传并发布，本平台仅提供信息存储服务。如涉及版权问题，请联系我们并提供版权证明，我们将立即删除！
