pages/index.md

Repro - Provably Data-driven Multiple Hyper-parameter Tuning with Structured Loss Function

Pages

Page
Claim 1: Establishes first general framework for generalization guarantees for multi-dimensional hyperparameter tuning
Claim 2: Strengthens guarantees using semi-algebraic function classes with finite fat-shattering dimension
Claim 3: Derives improved bounds for data-driven weighted group lasso and weighted fused lasso
Conclusion