Manifold constrained steepest descent for smooth and closed-set optimization
流形约束最速下降
机构 * Department of Mathematics, University of Hong Kong, Hong Kong, China(香港大学数学系)
AI总结 本文提出流形约束最速下降(MCSD)方法,通过单循环框架在流形上进行优化,并引入SPEL在Stiefel流形上的应用,实现了更稳定的性能和可扩展性。
Comments 29 pages, 3 figures, and 2 tables. Compared to the previous version, introduced a new variant MCSD-TP and a hybrid algorithm MCSD-PGD for close-set optimization