Distilling a Neural Network Into a Soft Decision Tree
arXiv:1711.09784
Abstract
Deep neural networks have proved to be a very effective way to perform classification tasks. They excel when the input data is high dimensional, the relationship between the input and the output is complicated, and the number of labeled training examples is large. But it is hard to explain why a learned network makes a particular classification decision on a particular test case. This is due to their reliance on distributed hierarchical representations. If we could take the knowledge acquired by the neural net and express the same knowledge in a model that relies on hierarchical decisions instead, explaining a particular decision would be much easier. We describe a way of using a trained neural net to create a type of soft decision tree that generalizes better than one learned directly from the training data.
presented at the CEX workshop at AI*IA 2017 conference
References in corpus (3)
Cited by in corpus (112)
- Interpretable machine learning: definitions, methods, and applications
- A trans-disciplinary review of deep learning research for water resources scientists
- Deep Neural Networks and Tabular Data: A Survey
- A Survey on the Explainability of Supervised Machine Learning
- Interpretable Machine Learning -- A Brief History, State-of-the-Art and Challenges
- Born Again Neural Networks
- FairSight: Visual Analytics for Fairness in Decision Making
- Interpreting Blackbox Models via Model Extraction
- Indoor Scene Understanding in 2.5/3D for Autonomous Agents: A Survey
- Hierarchical interpretations for neural network predictions
- Benchmarking Deep Learning Interpretability in Time Series Predictions
- ExplAIn: Explanatory Artificial Intelligence for Diabetic Retinopathy Diagnosis
- Analyzing Human-Human Interactions: A Survey
- Explainable Artificial Intelligence for Process Mining: A General Overview and Application of a Novel Local Explanation Approach for Predictive Process Monitoring
- Machine Learning in a data-limited regime: Augmenting experiments with synthetic data uncovers order in crumpled sheets
- NBDT: Neural-Backed Decision Trees
- Testing and verification of neural-network-based safety-critical control software: A systematic literature review
- Understanding trained CNNs by indexing neuron selectivity
- Dynamic Neural Networks: A Survey
- Evaluating Explanation Without Ground Truth in Interpretable Machine Learning
- Considerations When Learning Additive Explanations for Black-Box Models
- Neural Network Attributions: A Causal Perspective
- Explainable Machine Learning with Prior Knowledge: An Overview
- White Box Methods for Explanations of Convolutional Neural Networks in Image Classification Tasks
- Model Distillation for Revenue Optimization: Interpretable Personalized Pricing
- Interpreting CNNs via Decision Trees
- Born-Again Tree Ensembles
- FEED: Feature-level Ensemble for Knowledge Distillation
- Explaining Deep Neural Networks
- Scalable Rule-Based Representation Learning for Interpretable Classification
- What does it mean to understand a neural network?
- pForest: In-Network Inference with Random Forests
- Global Explanations of Neural Networks: Mapping the Landscape of Predictions
- Interpret Federated Learning with Shapley Values
- Trepan Reloaded: A Knowledge-driven Approach to Explaining Artificial Neural Networks
- Semantics and explanation: why counterfactual explanations produce adversarial examples in deep neural networks
- Unbox the Black-box for the Medical Explainable AI via Multi-modal and Multi-centre Data Fusion: A Mini-Review, Two Showcases and Beyond
- Exact and Consistent Interpretation for Piecewise Linear Neural Networks: A Closed Form Solution
- Attention Convolutional Binary Neural Tree for Fine-Grained Visual Categorization
- Causal Interpretability for Machine Learning -- Problems, Methods and Evaluation
- Teacher-Student Compression with Generative Adversarial Networks
- Rectified Decision Trees: Towards Interpretability, Compression and Empirical Soundness
- Transferring Inductive Biases through Knowledge Distillation
- Verification of Recurrent Neural Networks Through Rule Extraction
- Interpretable Machine Learning: Moving From Mythos to Diagnostics
- Learning Interpretable Models with Causal Guarantees
- LIT: Block-wise Intermediate Representation Training for Model Compression
- On Relating 'Why?' and 'Why Not?' Explanations
- Hybrid Predictive Model: When an Interpretable Model Collaborates with a Black-box Model
- A Comparative Study of Rule Extraction for Recurrent Neural Networks
- CDT: Cascading Decision Trees for Explainable Reinforcement Learning
- Improving the Interpretability of Deep Neural Networks with Knowledge Distillation
- Distilling a Deep Neural Network into a Takagi-Sugeno-Kang Fuzzy Inference System
- Towards Explainable Deep Learning for Credit Lending: A Case Study
- The What, the Why, and the How of Artificial Explanations in Automated Decision-Making
- Explaining Deep Neural Networks using Unsupervised Clustering
- Interpreting Deep Neural Networks Through Variable Importance
- Soft Gradient Boosting Machine
- Risk-Sensitive Compact Decision Trees for Autonomous Execution in Presence of Simulated Market Response
- Incorporating Priors with Feature Attribution on Text Classification
- Neural-encoding Human Experts' Domain Knowledge to Warm Start Reinforcement Learning
- Deep CNNs for Peripheral Blood Cell Classification
- Regional Tree Regularization for Interpretability in Black Box Models
- A novel method for extracting interpretable knowledge from a spiking neural classifier with time-varying synaptic weights
- HSD-CNN: Hierarchically self decomposing CNN architecture using class specific filter sensitivity analysis
- A New Approach for Explainable Multiple Organ Annotation with Few Data
- Rearchitecting Classification Frameworks For Increased Robustness
- Explainable Multivariate Time Series Classification: A Deep Neural Network Which Learns To Attend To Important Variables As Well As Informative Time Intervals
- Layerwise Knowledge Extraction from Deep Convolutional Networks
- Explaining Neural Networks Semantically and Quantitatively
- Adaptive wavelet distillation from neural networks through interpretations
- A Causal Lens for Peeking into Black Box Predictive Models: Predictive Model Interpretation via Causal Attribution
- Defining and Quantifying the Emergence of Sparse Concepts in DNNs
- Accelerating Large Scale Knowledge Distillation via Dynamic Importance Sampling
- Synthesising Reinforcement Learning Policies through Set-Valued Inductive Rule Learning
- Rectified Decision Trees: Exploring the Landscape of Interpretable and Effective Machine Learning
- A Survey of Techniques All Classifiers Can Learn from Deep Networks: Models, Optimizations, and Regularization
- Coloring the Black Box: Visualizing neural network behavior with a self-introspective model
- Formal Methods with a Touch of Magic
- Argumentation for Explainable Scheduling (Full Paper with Proofs)
- Deep Transfer Learning with Ridge Regression
- Transparent Classification with Multilayer Logical Perceptrons and Random Binarization
- Interpretable Time-series Classification on Few-shot Samples
- Learning Accurate Decision Trees with Bandit Feedback via Quantized Gradient Descent
- Efficient Encrypted Inference on Ensembles of Decision Trees
- Neural-to-Tree Policy Distillation with Policy Improvement Criterion
- Designing Interpretable Approximations to Deep Reinforcement Learning
- Extracting Interpretable Concept-Based Decision Trees from CNNs
- BDD4BNN: A BDD-based Quantitative Analysis Framework for Binarized Neural Networks
- LioNets: Local Interpretation of Neural Networks through Penultimate Layer Decoding
- RDPD: Rich Data Helps Poor Data via Imitation
- Channel Pruning via Multi-Criteria based on Weight Dependency
- Extracting Optimal Explanations for Ensemble Trees via Logical Reasoning
- Online probabilistic label trees
- SONG: Self-Organizing Neural Graphs
- XAlgo: a Design Probe of Explaining Algorithms' Internal States via Question-Answering
- Back to Square One: Superhuman Performance in Chutes and Ladders Through Deep Neural Networks and Tree Search
- Games for Fairness and Interpretability
- Computational principles of intelligence: learning and reasoning with neural networks
- Exact and Consistent Interpretation of Piecewise Linear Models Hidden behind APIs: A Closed Form Solution
- Convex Polytope Trees
- Efficient Modelling Across Time of Human Actions and Interactions
- Weakly Supervised Recovery of Semantic Attributes
- Learning Multi-Layered GBDT Via Back Propagation
- Computing Class Hierarchies from Classifiers
- How to Explain Neural Networks: an Approximation Perspective
- Adaptive Bayesian Reticulum
- A New Framework for Machine Intelligence: Concepts and Prototype
- Interpretable Mixture Density Estimation by use of Differentiable Tree-module
- Input Invex Neural Network
- A Neural Tangent Kernel Perspective of Infinite Tree Ensembles
- Making CNNs Interpretable by Building Dynamic Sequential Decision Forests with Top-down Hierarchy Learning