Global Explanations for Image Classifiers (Student Abstract)

We hypothesize that deep network classifications of complex scenes can be explained using sets of relevant objects. We employ beam search and singular value decomposition to generate local and global explanations that summarize the deep model's interpretation of a class.

Authors

• Bhavan K. Vasu Oregon State University
• Prasad Tadepalli Oregon State University

Keywords

• Deep Convolution Neural Network
• Deep Neural Network
• Explainable Artificial Intelligence
• Interpretable Machine Learning
• Model Interpretability
• Scene Understanding