Surrogate Losses for Online Learning of Stepsizes in Stochastic Non-Convex Optimization.
Zhenxun ZhuangAshok CutkoskyFrancesco OrabonaPublished in: ICML (2019)
Keyphrases
- convex optimization
- online learning
- regret bounds
- interior point methods
- online convex optimization
- e learning
- approximate dynamic programming
- low rank
- total variation
- convex optimization problems
- norm minimization
- primal dual
- convex relaxation
- active learning
- augmented lagrangian
- reinforcement learning
- linear combination
- low rank matrix
- convex formulation
- basis pursuit
- lower bound
- semidefinite program
- alternating direction method of multipliers