Spectral risk-based learning using unbounded losses

2021/05/11 by Holland, Matthew J., Haress, El Mehdi
#FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning (stat.ML)

paper · doi:10.48550/arxiv.2105.04816

Abstract

In this work, we consider the setting of learning problems under a wide class of spectral risk (or "L-risk") functions, where a Lipschitz-continuous spectral density is used to flexibly assign weight to extreme loss values. We obtain excess risk guarantees for a derivative-free learning procedure under unbounded heavy-tailed loss distributions, and propose a computationally efficient implementation which empirically outperforms traditional risk minimizers in terms of balancing spectral risk and misclassification error.

Related

• Supervised Learning with General Risk Functionals
• SOREL: A Stochastic Algorithm for Spectral Risks Minimization
• Stochastic Optimization for Spectral Risk Measures
• Learning without Concentration for General Loss Functions
• Excess Risk Bounds for Exponentially Concave Losses