A Survey on Neural Network Interpretability

Abstract

Along with the great success of deep neural networks, there is also growing concern about their black-box nature. The interpretability issue affects people's trust on deep learning systems. It is also related to many ethical problems, e.g., algorithmic discrimination. Moreover, interpretability is a desired property for deep networks to become powerful tools in other research fields, e.g., drug discovery and genomics. In this survey, we conduct a comprehensive review of the neural network interpretability research. We first clarify the definition of interpretability as it has been used in many different contexts. Then we elaborate on the importance of interpretability and propose a novel taxonomy organized along three dimensions: type of engagement (passive vs. active interpretation approaches), the type of explanation, and the focus (from local to global interpretability). This taxonomy provides a meaningful 3D view of distribution of papers from the relevant literature as two of the dimensions are not simply categorical but allow ordinal subcategories. Finally, we summarize the existing interpretability evaluation methods and suggest possible research directions inspired by our new taxonomy.

arXiv commentThis work has been accepted by IEEE-TETCI

References in corpus (39)

• Distributed Representations of Words and Phrases and their Compositionality
• Sequence to Sequence Learning with Neural Networks
• A Unified Approach to Interpreting Model Predictions
• Google's Neural Machine Translation System: Bridging the Gap between Human and Machine Translation
• Towards A Rigorous Science of Interpretable Machine Learning
• Axiomatic Attribution for Deep Networks
• Methods for Interpreting and Understanding Deep Neural Networks
• Striving for Simplicity: The All Convolutional Net
• Learning Important Features Through Propagating Activation Differences
• European Union regulations on algorithmic decision-making and a "right to explanation"
• Understanding Neural Networks Through Deep Visualization
• Understanding Black-box Predictions via Influence Functions
• Searching for Exotic Particles in High-Energy Physics with Deep Learning
• Unmasking Clever Hans Predictors and Assessing What Machines Really Learn
• Interpretable Explanations of Black Boxes by Meaningful Perturbation
• How to Explain Individual Classification Decisions
• The Loss Surfaces of Multilayer Networks
• Interpretable Machine Learning -- A Brief History, State-of-the-Art and Challenges
• Sanity Checks for Saliency Maps
• Towards Explainable Artificial Intelligence
• GANSpace: Discovering Interpretable GAN Controls
• Visualizing Deep Neural Network Decisions: Prediction Difference Analysis
• A Benchmark for Interpretability Methods in Deep Neural Networks
• Prototype selection for interpretable classification
• GAN Lab: Understanding Complex Deep Generative Models using Interactive Visual Experimentation
• Unsupervised Discovery of Interpretable Directions in the GAN Latent Space
• Deep Learning for Case-Based Reasoning through Prototypes: A Neural Network that Explains Its Predictions
• Model Agnostic Supervised Local Explanations
• Fooling Neural Network Interpretations via Adversarial Model Manipulation
• Beyond Sparsity: Tree Regularization of Deep Models for Interpretability
• Deep Learning of Quasar Spectra to Discover and Characterize Damped Lya Systems
• Mathematics of Deep Learning
• Feature Importance Ranking for Deep Learning
• Controlling generative models with continuous factors of variations
• Representer Point Selection for Explaining Deep Neural Networks
• Robustness of Conditional GANs to Noisy Labels
• Regularizing Black-box Models for Improved Interpretability
• On the Implicit Bias of Dropout
• How Robust are Deep Neural Networks?

Cited by in corpus (14)

• A Survey on Data-driven Software Vulnerability Assessment and Prioritization
• On Information Plane Analyses of Neural Network Classifiers -- A Review
• Deep Weakly-Supervised Learning Methods for Classification and Localization in Histology Images: A Survey
• Synthetic Benchmarks for Scientific Research in Explainable Machine Learning
• DuTrust: A Sentiment Analysis Dataset for Trustworthiness Evaluation
• Ensembles of Random SHAPs
• Interpretable Neural Networks with Frank-Wolfe: Sparse Relevance Maps and Relevance Orderings
• Unsupervised Learning of Neurosymbolic Encoders
• Image recognition via Vietoris-Rips complex
• Interpretability in Safety-Critical FinancialTrading Systems
• Interpretable Face Manipulation Detection via Feature Whitening
• How to Explain Neural Networks: an Approximation Perspective
• Tensorization of neural networks for improved privacy and interpretability
• A Robust Unsupervised Ensemble of Feature-Based Explanations using Restricted Boltzmann Machines