Corpus ID: 220128073

PAC-Bayesian Bound for the Conditional Value at Risk

@article{Mhammedi2020PACBayesianBF,
  title={PAC-Bayesian Bound for the Conditional Value at Risk},
  author={Zakaria Mhammedi and Benjamin Guedj and R. C. Williamson},
  journal={ArXiv},
  year={2020},
  volume={abs/2006.14763}
}
Conditional Value at Risk (CVaR) is a family of "coherent risk measures" which generalize the traditional mathematical expectation. Widely used in mathematical finance, it is garnering increasing interest in machine learning, e.g., as an alternate approach to regularization, and as a means for ensuring fairness. This paper presents a generalization bound for learning algorithms that minimize the CVaR of the empirical loss. The bound is of PAC-Bayesian type and is guaranteed to be small when the… Expand
4 Citations
Adaptive Sampling for Stochastic Risk-Averse Learning
  • 9
  • PDF
A PAC-Bayesian Perspective on Structured Prediction with Implicit Loss Embeddings
  • 1
  • PDF
Upper and Lower Bounds on the Performance of Kernel PCA
  • PDF

References

SHOWING 1-10 OF 52 REFERENCES
Concentration bounds for empirical conditional value-at-risk: The unbounded case
  • 18
  • PDF
Optimizing the CVaR via Sampling
  • 87
  • PDF
Concentration Inequalities for Conditional Value at Risk
  • 9
  • PDF
Entropic Value-at-Risk: A New Coherent Risk Measure
  • 134
  • Highly Influential
Concentration of risk measures: A Wasserstein distance approach
  • 12
  • Highly Influential
  • PDF
Adaptive Sampling for Stochastic Risk-Averse Learning
  • 9
  • PDF
PAC-Bayes Un-Expected Bernstein Inequality
  • 7
  • PDF
A robust approach based on conditional value-at-risk measure to statistical learning problems
  • 30
  • PDF
Variance-based Regularization with Convex Objectives
  • 124
  • PDF
Algorithms for CVaR Optimization in MDPs
  • 87
  • PDF
...
1
2
3
4
5
...