GTC 2020: Scaling Deep Learning Interpretability by Visualizing Activation and Attribution Summarizations
After clicking “Watch Now” you will be prompted to login or join.
Click “Watch Now” to login or join the NVIDIA Developer Program.
Scaling Deep Learning Interpretability by Visualizing Activation and Attribution Summarizations
Polo Chau, The Georgia Institute of Technology | Fred Hohman, Georgia Tech
We'll present Summit, an interactive system that scalably and systematically summarizes and visualizes the features that a deep learning model has learned, and how those features interact to make predictions. Summit introduces two new scalable summarization techniques: activation aggregation discovers important neurons, and neuron-influence aggregation identifies relationships among such neurons. Summit combines these techniques to create the novel attribution graph that reveals and summarizes crucial neuron associations and substructures that contribute to a model's outcomes. Summit scales to large data, such as the ImageNet dataset with 1.2 million images, and leverages neural network feature visualization and dataset examples to help users distill large, complex neural network models into compact, interactive visualizations. We'll present neural network exploration scenarios where Summit helps us discover multiple surprising insights into a prevalent, large-scale image classifier's learned representations and informs future neural network architecture design. The Summit visualization runs in modern web browsers and is open sourced.