报告题目 (Title)：Descent-Net: Learning Descent Directions for Constrained Optimization（Descent-Net：面向约束优化的下降方向学习网络）
报告人 (Speaker)：陈士祥 研究员（中国科学技术大学）
报告时间 (Time)：2025年10月31日 (周五) 14:30
报告地点 (Place)：校本部F309
邀请人(Inviter)：徐姿 教授
报告摘要：Deep learning approaches, known for their ability to model complex relationships and fast execution, are increasingly being applied to solve large optimization problems. However, existing methods often face challenges in simultaneously ensuring feasibility and achieving an optimal objective value. To address this issue, we propose Descent-Net, a neural network designed to learn an effective descent direction from a feasible solution. By updating the solution along this learned direction, Descent-Net improves the objective value while preserving feasibility. We also provide theoretical advantages of the proposed algorithm. Our method demonstrates strong performance on both synthetic optimization tasks and the real-world AC optimal power flow problem.