Beyond Word Importance: Contextual Decomposition for Interpreting LSTMs
About this video
Hand-picked by a BrighterTube curator · found via round-3 back catalogue of a channel Lumen picks video by video · 82% confidenceCovers contextual decomposition methods for interpreting LSTM behavior in NLP tasks. Teaches techniques for understanding which input features drive specific neural network predictions.
A 18-minute video from Berkeley Institute for Data Science (BIDS): “Beyond Word Importance: Contextual Decomposition for Interpreting LSTMs”.
Video by Berkeley Institute for Data Science (BIDS), played through YouTube's embedded player.
Part of BIDS TextXD 2017