Shuai Zhang, 博士毕业于伦斯勒理工学院(RPI), 本科毕业于中国科学技术大学(USTC), 于2023年秋季加入新泽西理工学院(NJIT)数据科学系担任助理教授。他的研究目标主要在于theoretical foundations of AI和design of principled and efficient learning algorithms,以实现更好,更安全和更高效的AI learning systems。目前的兴趣在于大模型中的迁移学习(parameter-efficient transfer learning)和稀疏学习(sparse learning)。他与IBM Thomas J. Watson Research Center和MIT-IBM Watson AI Lab有过深入的合作经历。他作为第一作者在多个著名ML/AI/信号处理期刊会议 ( ICML / ICLR / NeurIPS/ TSP / TNNLS )发表论文。
-Learning Theory: generalization ability of deep learning model & algorithms.
-Non-convex Optimization: convergence analysis of training emerging neural networks, convergence analysis of other non-convex machine learning problems.
-Parameter-efficient Transfer Learning: focusing on both the theoretical foundations and algorithm design of parameter-efficient transfer learning approach, especially in the applications involved foundation models (Large Language Models).
-Deep Reinforcement Learning: Transfer learning and RL from human feedback
-Graph Neural Network: graph sampling & sparsification in GNN learning.
-Spatial-temporal Data Analysis.
招生: