Semantics for Global and Local Interpretation of Deep Neural Networks

2019/10/21 by Jindong Gu, Volker Tresp, Gu, Jindong +1
Computer Science · #Computer Vision and Pattern Recognition (cs.CV) #Domain Adaptation and Few-Shot Learning #Explainable Artificial Intelligence (XAI) #FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning and Data Classification

paper · pdf · doi:10.48550/arxiv.1910.09085

openalex publication_date 2019/10/21 · openalex created_date 2025/10/10 · openalex updated_date 2026/08/04

Abstract

Deep neural networks (DNNs) with high expressiveness have achieved state-of-the-art performance in many tasks. However, their distributed feature representations are difficult to interpret semantically. In this work, human-interpretable semantic concepts are associated with vectors in feature space. The association process is mathematically formulated as an optimization problem. The semantic vectors obtained from the optimal solution are applied to interpret deep neural networks globally and locally. The global interpretations are useful to understand the knowledge learned by DNNs. The interpretation of local behaviors can help to understand individual decisions made by DNNs better. The empirical experiments demonstrate how to use identified semantics to interpret the existing DNNs.

Citations

• Understanding Intra-Class Knowledge Inside CNN
• Object Detectors Emerge in Deep Scene CNNs
• ImageNet Large Scale Visual Recognition Challenge
• DeCAF: A Deep Convolutional Activation Feature for Generic Visual\n Recognition
• ImageNet classification with deep convolutional neural networks
• Multifaceted Feature Visualization: Uncovering the Different Types of\n Features Learned By Each Neuron in Deep Neural Networks
• Axiomatic Attribution for Deep Networks
• SmoothGrad: removing noise by adding noise
• The (Un)reliability of saliency methods
• Sanity Checks for Saliency Maps
• Deep Inside Convolutional Networks: Visualising Image Classification Models and Saliency Maps

Related

• Choosing the right basis for interpretability: Psychophysical comparison between neuron-based and dictionary-based representations
• An Information Theoretic Interpretation to Deep Neural Networks
• Locally Testing Model Detections for Semantic Global Concepts
• Improving Interpretability of Deep Neural Networks with Semantic Information
• Local Explanation Methods for Deep Neural Networks Lack Sensitivity to Parameter Values