Extracting decision trees from trained neural networks

By Olcay Boz. 2002

154 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Information Systems, Safety, Robustness & Interpretability

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Information Systems

In Safety, Robustness & Interpretability

Citations in the corpus

Local Rule-Based Explanations of Black Box Decision Systems (2018, methodology): “Decision trees have been adopted to globally explain neural networks [9, 22] and tree ensembles [18, 34].”
A Survey of Methods for Explaining Black Box Models (2018, methodology): “DT NN TAB X X X DecText [9] Boz 2002 Model Expl.”
Explaining nonlinear classification decisions with deep Taylor decomposition (2015, background): “transformations of the data [31], the role of particular neurons [32], or its global decision structure [33, 34].”
Toward Interpretable Deep Reinforcement Learning with Linear Model U-Trees (2018, methodology): “art Velocity 8087.68 Cart Position 7171.71 Pole Velocity At Tip 2953.73 Fig.6: Super pixels in Flappy Bird 6.2 Rule Extraction Rule extraction is a common method to extract knowledge from tree models [7,6,2]. We extract and analyze rules for the Mountain Car and Cart Pole environment. Figure 7 (top)”
Causal Interpretability for Machine Learning - Problems, Methods and Evaluation (2020, methodology): “Inspired by TREPAN, Boz [10] propose a method called DECTEXT to extract a decision tree that mimics the behavior of a trained Neural Network.”
Explaining Predictions of Non-Linear Classifiers in NLP (2016, background): “, 2016a) or through decision tree extraction (Krishnan et al., 1999).”
Open the Black Box Data-Driven Explanation of Black Box Decision Systems (2018, methodology): “For example, decision trees have been adopted to explain neural networks [26] and tree ensembles [22, 42], while decision rules have been widely used to explain neural networks [1, 2] and support vector machines [16].”
Benchmarking and survey of explanation methods for black box models (2021, background): “DecText is a global model-specific explainer tailored for neural networks (Boz 2002).”
NBDT: Neural-Backed Decision Trees (2020, background): “Work converting from decision trees to neural networks also dates back three decades [14,17,6,7].”
Counterfactual Explanations and Algorithmic Recourses for Machine Learning: A Review (2020, unknown): “Various papers have developed techniques to explain neural networks and tree ensembles using single decision tree [65, 83, 184] and rule sets [14, 76].”
Interpreting Shared Deep Learning Models via Explicable Boundary Trees (2017, methodology): “Another common approach is to mimic a complicated model using an interpretable simple model such as decision tree (Craven and Shavlik 1996; Schmitz, Aldrich, and Gouws 1999; Boz 2002; Che et al. 2016).”
Hybrid decision tree (2002, methodology): “Li et al. (1992) proposed a competitive neural network named adaptive neural tree whose nodes were organized in a tree topology for classification and vector quantization by combining competitive learning principles with structural adaptation during learning. Kubat (1998) initialized a RBF network w”
Explainable Artificial Intelligence: a Systematic Review (2020, methodology): “The DecText method [249] also extracts high fidelity DTs from a DNN.”
Counterfactual Explanations for Machine Learning: A Review (2020, background): “Various papers have developed techniques to explain neural networks and tree ensembles using single decision tree [25, 70, 34] and rule sets [28, 13].”
On Interpretability of Artificial Neural Networks: A Survey (2020, methodology): “The gist of direct extraction is to construct a new interpretable model such as a decision tree [92], [192] or a rule-based system directly from the trained model.”
A Survey on the Explainability of Supervised Machine Learning (2020, methodology): “(2007a) cAntMinerPB Classification Rule-based Otero and Freitas (2016) CART Classification & Regression Tree-based Breiman (2017) ID2of3 Classification & Regression Tree-based Craven and Shavlik (1996) ID3 Classification & Regression Tree-based Quinlan (1986) C4.”
Neural Decision Trees (2017, methodology): “Using an ANN to ﬁlter a dataset prior to learning a DT is explored in Krishnan et al. (1999).”