Distributionally Robust Regularization of Sparse Integer Programming Trained Learning Models

Name: Distributionally Robust Regularization of Sparse Integer Programming Trained Learning Models
Uploaded: 2026-10-17T17:06:01.717895+00:00
Duration: 505 s
Description: The paper proposes a method to build explainable binary classification models by learning sparse sets of feature value rules. It uses Distributionally Robust Optimization (DRO) instead of expensive cross-validation to select the best rule set combination, balancing model accuracy with sparsity and ensuring strong generalization guarantees.

The paper proposes a method to build explainable binary classification models by learning sparse sets of feature value rules. It uses Distributionally Robust Optimization (DRO) instead of expensive cross-validation to select the best rule set combination, balancing model accuracy with sparsity and ensuring strong generalization guarantees.

Open the full episode