Shypd.aiThe largest open index of AI toolsDiCE is an open-source Python library for generating diverse counterfactual explanations for machine learning models. It helps users understand how changes to input features can affect model predictions and provides actionable insights.
Trending
About
DiCE (Diverse Counterfactual Explanations) is an open-source Python library designed to generate counterfactual explanations for any machine learning model. It addresses the critical need for interpretability in ML, especially in sensitive domains like finance and healthcare. Unlike traditional explanations that might only state why a decision was made (e.g., "poor credit history"), DiCE provides "what-if" scenarios, showing minimal changes to input features that would alter a model's prediction (e.g., "you would have received the loan if your income was higher by $10,000"). It supports various methods for generating counterfactuals, including model-agnostic approaches like randomized sampling and genetic algorithms, as well as gradient-based methods for deep learning models. DiCE allows users to tune parameters for diversity and proximity of explanations, specify feature weights to reflect difficulty of change, and define constraints on features for practical feasibility. It is a valuable tool for ML model developers to debug models and for decision subjects to understand actionable recourse.
Capabilities
Pricing & Plans
Open Source
Free
FAQs
Trending
