Making Tree Ensembles Interpretable
arXiv:1606.05390
Abstract
Tree ensembles, such as random forest and boosted trees, are renowned for their high prediction performance, whereas their interpretability is critically limited. In this paper, we propose a post processing method that improves the model interpretability of tree ensembles. After learning a complex tree ensembles in a standard way, we approximate it by a simpler model that is interpretable for human. To obtain the simpler model, we derive the EM algorithm minimizing the KL divergence from the complex ensemble. A synthetic experiment showed that a complicated tree ensemble was approximated reasonably as interpretable.
arXiv commentpresented at 2016 ICML Workshop on Human Interpretability in Machine Learning (WHI 2016), New York, NY
Cited by in corpus (17)
- A Survey on the Explainability of Supervised Machine Learning
- Local Rule-Based Explanations of Black Box Decision Systems
- Concise Fuzzy System Modeling Integrating Soft Subspace Clustering and Sparse Learning
- A One-Class Classification Decision Tree Based on Kernel Density Estimation
- Tree Space Prototypes: Another Look at Making Tree Ensembles Interpretable
- Open the Black Box Data-Driven Explanation of Black Box Decision Systems
- Scalable Rule-Based Representation Learning for Interpretable Classification
- Instance-Level Explanations for Fraud Detection: A Case Study
- Making Tree Ensembles Interpretable: A Bayesian Model Selection Approach
- An exact counterfactual-example-based approach to tree-ensemble models interpretability
- Silas: High Performance, Explainable and Verifiable Machine Learning
- Technologies for Trustworthy Machine Learning: A Survey in a Socio-Technical Context
- Reasons, Values, Stakeholders: A Philosophical Framework for Explainable Artificial Intelligence
- Making Logic Learnable With Neural Networks
- e-QRAQ: A Multi-turn Reasoning Dataset and Simulator with Explanations
- Explaining Anomalies in Groups with Characterizing Subspace Rules
- Interpretable Summaries of Black Box Incident Triaging with Subgroup Discovery