Now You See Me (CME): Concept-based Model Extraction

Abstract

Deep Neural Networks (DNNs) have achieved remarkable performance on a range of tasks. A key step to further empowering DNN-based approaches is improving their explainability. In this work we present CME: a concept-based model extraction framework, used for analysing DNN models via concept-based extracted models. Using two case studies (dSprites, and Caltech UCSD Birds), we demonstrate how CME can be used to (i) analyse the concept information learned by a DNN model (ii) analyse how a DNN uses this concept information when predicting output labels (iii) identify key concept information that can further improve DNN predictive performance (for one of the case studies, we showed how model accuracy can be improved by over 14%, using only 30% of the available concepts).

Presented at the AIMLAI workshop at the 29th ACM International Conference on Information and Knowledge Management (CIKM 2020)

References in corpus (8)

• Scikit-learn: Machine Learning in Python
• European Union regulations on algorithmic decision-making and a "right to explanation"
• Object Detectors Emerge in Deep Scene CNNs
• Sanity Checks for Saliency Maps
• Towards Robust Interpretability with Self-Explaining Neural Networks
• Concept Bottleneck Models
• Explaining Classifiers with Causal Concept Effect (CaCE)
• On Completeness-aware Concept-Based Explanations in Deep Neural Networks

Cited by in corpus (7)

• Logic Explained Networks
• Entropy-based Logic Explanations of Neural Networks
• Towards Robust Metrics for Concept Representation Evaluation
• Statistically Significant Concept-based Explanation of Image Classifiers via Model Knockoffs
• Towards Human-Interpretable Prototypes for Visual Assessment of Image Classification Models
• PyTorch, Explain! A Python library for Logic Explained Networks
• A Framework to Learn with Interpretation