Convex Loss Function

Convex loss functions are fundamental in machine learning, enabling efficient optimization algorithms to find model parameters minimizing prediction error. Current research focuses on extending their application to distributed and federated learning settings, addressing challenges like data heterogeneity, adversarial attacks, and stragglers, often employing techniques like mirror descent and gradient coding. This work also explores tighter convex loss functions (e.g., Fitzpatrick losses) and their application in robust optimization frameworks, including distributionally robust optimization and differentially private optimization, to improve model robustness and privacy. The resulting advancements enhance the efficiency, robustness, and privacy of machine learning models across various applications.

Papers

September 2, 2022

Optimizing the Performative Risk under Weak Convexity Assumptions
Yulai Zhao
Supervised Learning Performative Prediction Strong Convexity Convex Loss Function Minimization Problem

April 3, 2022

A Differentially Private Framework for Deep Learning with Convexified Loss Functions
Zhigang Lu, Hassan Jameel Asghar, Mohamed Ali Kaafar, Darren Webb, Peter Dickinson
Deep Learning Differential Privacy Private Stochastic Differential Privacy SGD Convex Loss Function Agentive Permission

December 8, 2021

The perils of being unhinged: On the accuracy of classifiers minimizing a noise-robust convex loss
Philip M. Long, Rocco A. Servedio
Loss Function Label Noise Simple Classifier Binary Classifier Convex Loss Function Noise Robust Loss

October 14, 2021

Model-Change Active Learning in Graph-Based Semi-Supervised Learning
Kevin Miller, Andrea L. Bertozzi
LeArning Abstract Graph Drawing Active Learning Semi Supervised Learning Semi Supervised Classification Convex Loss Function

February 9, 2021

Output Perturbation for Differentially Private Convex Optimization: Faster and More General
Andrew Lowy, Meisam Razaviyayn
Private Algorithm Private Gradient Convex Loss Function Private Convex Optimization Private Empirical Risk Minimization

Convex Loss Function

Papers

Optimizing the Performative Risk under Weak Convexity Assumptions

A Differentially Private Framework for Deep Learning with Convexified Loss Functions

The perils of being unhinged: On the accuracy of classifiers minimizing a noise-robust convex loss

Model-Change Active Learning in Graph-Based Semi-Supervised Learning

Output Perturbation for Differentially Private Convex Optimization: Faster and More General