Stochastic Convex Optimization

Stochastic convex optimization focuses on efficiently finding the minimum of a convex function whose value is only known through noisy samples, a common problem in machine learning. Current research emphasizes developing algorithms with optimal convergence rates under various constraints, including differential privacy requirements, heavy-tailed data, and limited computational resources; adaptive gradient methods and variance reduction techniques are prominent approaches. These advancements are crucial for improving the scalability and reliability of machine learning models while addressing privacy concerns and handling real-world data complexities.

Papers

April 19, 2022

Making Progress Based on False Discoveries
Roi Livni
Gradient Descent Much Progress Gradient Estimation Stochastic Convex Optimization Noisy Evaluation

March 22, 2022

Provable Constrained Stochastic Convex Optimization with XOR-Projected Gradient Descent
Fan Ding, Yijie Wang, Jianzhu Ma, Yexiang Xue
Gradient Descent Stochastic Convex Optimization Gradient Episodic Memory

March 2, 2022

Efficient Online Linear Control with Stochastic Convex Costs and Unknown Dynamics
Asaf Cassel, Alon Cohen, Tomer Koren
Dynamical System Stochastic Convex Optimization Unknown Dynamic Parameter Uncertainty Regret Rate Linear Control

February 23, 2022

Mirror Descent Strikes Again: Optimal Stochastic Convex Optimization under Infinite Noise Variance
Nuri Mert Vural, Lu Yu, Krishnakumar Balasubramanian, Stanislav Volgushev, Murat A. Erdogdu
Stochastic Gradient Stochastic Approximation Mirror Descent Stochastic Convex Optimization Stochastic Mirror Descent Unbounded Variance Mirror Map

December 31, 2021

Stochastic convex optimization for provably efficient apprenticeship learning
Angeliki Kamoutsi, Goran Banjac, John Lygeros
Reinforcement Learning Imitation Learning Inverse Reinforcement Learning Stochastic Convex Optimization Apprenticeship Learning

December 28, 2021

Efficient Performance Bounds for Primal-Dual Reinforcement Learning from Demonstrations
Angeliki Kamoutsi, Goran Banjac, John Lygeros
Inverse Reinforcement Learning Noisy Demonstration Policy Optimization Stochastic Convex Optimization Low Regret Performance Bound Primal Dual Reinforcement Learning

Stochastic Convex Optimization

Papers

Making Progress Based on False Discoveries

Provable Constrained Stochastic Convex Optimization with XOR-Projected Gradient Descent

Efficient Online Linear Control with Stochastic Convex Costs and Unknown Dynamics

Stability vs Implicit Bias of Gradient Methods on Separable Data and Beyond

Benign Underfitting of Stochastic Gradient Descent

Thinking Outside the Ball: Optimal Learning with Gradient Descent for Generalized Linear Stochastic Convex Optimization

Mirror Descent Strikes Again: Optimal Stochastic Convex Optimization under Infinite Noise Variance

Stochastic convex optimization for provably efficient apprenticeship learning

Efficient Performance Bounds for Primal-Dual Reinforcement Learning from Demonstrations