Causal Learning and Explanation of Deep Neural Networks via Autoencoded Activations
arXiv:1802.00541
Abstract
Deep neural networks are complex and opaque. As they enter application in a variety of important and safety critical domains, users seek methods to explain their output predictions. We develop an approach to explaining deep neural networks by constructing causal models on salient concepts contained in a CNN. We develop methods to extract salient concepts throughout a target network by using autoencoders trained to extract human-understandable representations of network activations. We then build a bayesian causal model using these extracted concepts as variables in order to explain image classification. Finally, we use this causal model to identify and visualize features with significant causal influence on final classification.
References in corpus (3)
Cited by in corpus (14)
- Responsible and Regulatory Conform Machine Learning for Medicine: A Survey of Challenges and Solutions
- Inspect, Understand, Overcome: A Survey of Practical Methods for AI Safety
- Causality Learning: A New Perspective for Interpretable Machine Learning
- Explaining Deep Learning Models using Causal Inference
- When Causal Intervention Meets Adversarial Examples and Image Masking for Deep Neural Networks
- Generative causal explanations of black-box classifiers
- Foundations of Explainable Knowledge-Enabled Systems
- Causal Interpretability for Machine Learning -- Problems, Methods and Evaluation
- Adversarial Attacks and Defenses: An Interpretation Perspective
- Learning Latent Causal Structures with a Redundant Input Neural Network
- Defining and Quantifying the Emergence of Sparse Concepts in DNNs
- Explainability-aided Domain Generalization for Image Classification
- Examining the causal structures of deep neural networks using information theory
- From Demonstrations to Task-Space Specifications: Using Causal Analysis to Extract Rule Parameterization from Demonstrations