An overview of visualization techniques used for computer vision (CNNs) and natural language processing (RNNs), as well as generic feature representation methods for explainability.
-
CNN Visualization - Computer Vision (images)
Sanity Checks for Saliency Maps -
RNN Visualization - Natural Language Processing (text)
LSTMVis: A Tool for Visual Analysis of Hidden State Dynamics in Recurrent Neural Networks -
Explainability (features)
A Unified Approach to Interpreting Model Predictions