报告题目:Near-Optimal Convex Simple Bilevel Optimization with a Bisection Method
报告人:江如俊 副教授(复旦大学)
主持人:王祥丰 副教授
报告时间:2024年3月18日(星期一)13:00-14:00
报告地点:华东师范大学普陀校区理科楼B112
报告摘要:
This talk focuses on a class of simple bilevel optimization problems where we minimize a composite convex function at the upper-level subject to a composite convex lower-level problem. Existing methods either provide asymptotic guarantees for the upper-level objective or attain slow sublinear convergence rates. We propose a bisection algorithm to find an approximate solution. In each iteration, the binary search narrows the interval by assessing inequality system feasibility. Under mild conditions, we show that our method achieves near-optimal rates, matching that in unconstrained smooth or composite convex optimization when disregarding logarithmic terms. Numerical experiments demonstrate the effectiveness of our method.
报告人简介:
江如俊,复旦大学大数据学院副教授,博士生导师。2016年7月于香港中文大学获得博士学位。研究方向主要包括优化算法和理论分析,二次规划,及其在运筹学、机器学习和金融工程领域的应用。其研究成果发表在Math. Program., SIAM J. Optim.、Math. Oper. Res.、INFORMS J. Comput.和ICML、NeurIPS等国际顶级期刊或会议上。获上海市扬帆计划、国家级青年人才计划支持,主持国家自然科学基金青年项目和面上项目。获国际机器学习大会ICML 2022杰出论文奖。