Adaptive Stochastic Gradient Descents on Manifolds with an Application on Weighted Low-Rank Approximation

This work addresses optimization challenges in machine learning for researchers and practitioners dealing with manifold-based problems, though it appears incremental as it extends existing methods with adaptive learning rates.

The paper tackles the problem of optimizing stochastic gradient descent on manifolds by proving a convergence theorem for adaptive learning rates, and applies this to achieve weighted low-rank approximation with improved convergence guarantees.