New SCENT algorithm improves optimization for entropic risk minimization

Researchers have developed a new algorithm called SCENT for compositional entropic risk minimization, a problem formulation involving Log-Expectation-Exponential functions. Existing methods for this type of optimization suffer from issues like non-convergence and slow convergence rates. SCENT utilizes a stochastic proximal mirror descent update, which the researchers claim offers theoretical advantages over standard SGD. The algorithm has demonstrated empirical effectiveness in applications such as extreme classification, partial AUC maximization, contrastive learning, and distributionally robust optimization, outperforming existing baselines. AI

IMPACT Introduces a more efficient optimization algorithm for machine learning tasks involving complex risk formulations.

RANK_REASON The cluster contains a research paper detailing a new algorithm and its theoretical and empirical evaluation. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• contrastive learning
• distributionally robust optimization
• extreme classification
• Log-Expectation-Exponential
• Log-Sum-Exponential
• SGD
• stochastic proximal mirror descent
• Xiyuan Wei

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →