A Random Walk Approach to First-Order Stochastic Convex Optimization

Date July 7, 2019
Authors Sattar Vakili, Qing Zhao (Cornell University)

An active search strategy based on devising a biased random walk on an infinite-depth tree constructed through successive partitioning of the search domain is developed. By localizing data processing to small subsets of the input domain based on the tree structure, it enjoys very low computation and memory complexity and allows dynamic allocation of limited data storage.

View the paper‚Äč


Related articles

What is Stochastic Network Control?

Soft Q-Learning with Mutual-Information Regularization