Variance Reduction

Variance reduction techniques aim to improve the efficiency and stability of various machine learning and optimization algorithms by decreasing the variability in estimates of gradients or other key quantities. Current research focuses on applying these techniques to diverse areas, including meta-learning, reinforcement learning, federated learning, and Monte Carlo methods, often employing algorithms like stochastic variance reduced gradient (SVRG), control variates, and importance sampling. These advancements lead to faster convergence, improved sample efficiency, and more robust performance in various applications, ultimately impacting the scalability and reliability of machine learning models and simulations.

Papers

November 15, 2024

MARS: Unleashing the Power of Variance Reduction for Training Large Models
Huizhuo Yuan, Yifeng Liu, Shuang Wu, Xun Zhou, Quanquan Gu
Full Model Real Power Variance Reduction Gradient Method Large Scale Training

November 11, 2024

Robust Fine-tuning of Zero-shot Models via Variance Reduction
Beier Zhu, Jiequan Cui, Hanwang Zhang
Fine Tuned Model Variance Reduction Robust Fine Tuning Zero Shot Model

November 3, 2024

Conformal Risk Minimization with Variance Reduction
Sima Noorani, Orlando Romero, Nicolo Dal Fabbro, Hamed Hassani, George J. Pappas
Conformal Prediction Objective Function Variance Reduction Conformal Risk Control

October 11, 2024

Variance reduction combining pre-experiment and in-experiment data
Zhexiao Lin, Pablo Crespo
Preliminary Study Variance Reduction Relevant Covariates Online Experiment Experimental Data

October 8, 2024

Pathwise Gradient Variance Reduction with Control Variates in Variational Inference
Kenyon Ng, Susan Wei
Variational Inference Variance Reduction Control Variate Variance Reduction Technique Variance Reduced

October 2, 2024

Reducing Variance in Meta-Learning via Laplace Approximation for Regression Tasks
Alfredo Reichlin, Gustaf Tegnér, Miguel Vasco, Hang Yin, Mårten Björkman, Danica Kragic
Meta Learning Regression Task Variance Reduction Gradient Based Meta Learning Laplace Approximation

August 27, 2024

The Benefits of Balance: From Information Projections to Variance Reduction
Lang Liu, Ronak Mehta, Soumik Pal, Zaid Harchaoui
Contrastive Learning Complementary Benefit Variance Reduction Balancing Strategy Multidimensional Projection Representation Collapse Self Supervised Clustering Stochastic Operator

August 22, 2024

Variance reduction of diffusion model's gradients with Taylor approximation-based control variate
Paul Jeha, Will Grathwohl, Michael Riis Andersen, Carl Henrik Ek, Jes Frellsen
Diffusion Model Natural Gradient Variance Reduction Score Matching Better Optimizers Control Variate Score Based Model

July 23, 2024

STATE: A Robust ATE Estimator of Heavy-Tailed Metrics for Variance Reduction in Online Controlled Experiments
Hao Zhou, Kun Sun, Shaoming Li, Yangfeng Fan, Guibin Jiang, Jiaqi Zheng, Tao Li
Internal State Variance Reduction Robust Estimation Conditional Average Treatment Effect Online Experiment Variance Reduction Technique

June 23, 2024

Semi-Variance Reduction for Fair Federated Learning
Saber Malekmohammadi
Loss Function Variance Reduction FAIR Federated Learning Fair Federated Learning

June 1, 2024

Efficient Sign-Based Optimization: Accelerating Convergence via Variance Reduction
Wei Jiang, Sifan Yang, Wenhao Yang, Lijun Zhang
Gradient Descent Early Stage Convergence Stochastic Gradient Variance Reduction Sign Based

May 28, 2024

Double Variance Reduction: A Smoothing Trick for Composite Optimization Problems without First-Order Gradient
Hao Di, Haishan Ye, Yueling Zhang, Xiangyu Chang, Guang Dai, Ivor W. Tsang
Variance Reduction Composite Optimization Variance Reduction Technique First Order Gradient

May 14, 2024

Gradient Estimation and Variance Reduction in Stochastic and Deterministic Models
Ronan Keane
Stochastic Way Variance Reduction Optimization Algorithm Gradient Based Optimization Deterministic Algorithm Gradient Estimation Gradient Estimator Nonlinear Programming

May 9, 2024

Variance Control for Black Box Variational Inference Using The James-Stein Estimator
Dominic B. Dayta
Variational Inference Variance Reduction Black Box Variational Inference Rao Blackwellized

April 2, 2024

Variance-Reduced Policy Gradient Approaches for Infinite Horizon Average Reward Markov Decision Processes
Swetha Ganesh, Washim Uddin Mondal, Vaneet Aggarwal
Variance Reduction Simple Regret Variance Reduced Policy Gradient

March 6, 2024

Targeted Variance Reduction: Robust Bayesian Optimization of Black-Box Simulators with Noise Parameters
John Joshua Miller, Simon Mak
Optimization Purpose Robust Optimization Variance Reduction Noise Estimation Robust Performance Stochastic Robust Optimization Robust Bayesian Optimization

February 14, 2024

Variance Reduction and Low Sample Complexity in Stochastic Optimization via Proximal Point Method
Jiaming Liang
Stochastic Gradient Stochastic Optimization Variance Reduction Proximal Point Gradient Noise Convex Composite Optimization Stochastic Proximal Point

February 6, 2024

Averaging $n$-step Returns Reduces Variance in Reinforcement Learning
Brett Daley, Martha White, Marlos C. Machado
Reinforcement Learning Variance Reduction Return Trajectory

February 2, 2024

Why do Random Forests Work? Understanding Tree Ensembles as Self-Regularizing Adaptive Smoothers
Alicia Curth, Alan Jeffares, Mihaela van der Schaar
Random Forest Variance Reduction Tree Ensemble Hypothesis Class Learned Function

January 8, 2024

Variance Reduction in Ratio Metrics for Efficient Online Experiments
Shubham Baweja, Neeti Pokharna, Aleksei Ustimenko, Olivier Jeunen
Variance Reduction Online Experiment Control Variate Variance Reduction Technique B Test

Variance Reduction

Papers

MARS: Unleashing the Power of Variance Reduction for Training Large Models

Robust Fine-tuning of Zero-shot Models via Variance Reduction

Conformal Risk Minimization with Variance Reduction

Variance reduction combining pre-experiment and in-experiment data

Pathwise Gradient Variance Reduction with Control Variates in Variational Inference

Reducing Variance in Meta-Learning via Laplace Approximation for Regression Tasks

The Benefits of Balance: From Information Projections to Variance Reduction

Variance reduction of diffusion model's gradients with Taylor approximation-based control variate

STATE: A Robust ATE Estimator of Heavy-Tailed Metrics for Variance Reduction in Online Controlled Experiments

Semi-Variance Reduction for Fair Federated Learning

Efficient Sign-Based Optimization: Accelerating Convergence via Variance Reduction

Double Variance Reduction: A Smoothing Trick for Composite Optimization Problems without First-Order Gradient

Gradient Estimation and Variance Reduction in Stochastic and Deterministic Models

Variance Control for Black Box Variational Inference Using The James-Stein Estimator

Variance-Reduced Policy Gradient Approaches for Infinite Horizon Average Reward Markov Decision Processes

Targeted Variance Reduction: Robust Bayesian Optimization of Black-Box Simulators with Noise Parameters

Variance Reduction and Low Sample Complexity in Stochastic Optimization via Proximal Point Method

Averaging $n$-step Returns Reduces Variance in Reinforcement Learning

Why do Random Forests Work? Understanding Tree Ensembles as Self-Regularizing Adaptive Smoothers

Variance Reduction in Ratio Metrics for Efficient Online Experiments