English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Current techniques for explainable AI have been applied with some success toimage processing. The recent rise of research in video processing has calledfor similar work n deconstructing and explaining spatio-temporal models. Whilemany techniques are designed for 2D convolutional models, others are inherentlyapplicable to any input domain. One such body of work, deep Taylordecomposition, propagates relevance from the model output distributively ontoits input and thus is not restricted to image processing models. However, byexploiting a simple technique that removes motion information, we show that itis not the case that this technique is effective as-is for representingrelevance in non-image tasks. We instead propose a discriminative method thatproduces a na\"ive representation of both the spatial and temporal relevance ofa frame as two separate objects. This new discriminative relevance modelexposes relevance in the frame attributed to motion, that was previouslyambiguous in the original explanation. We observe the effectiveness of thistechnique on a range of samples from the UCF-101 action recognition dataset,two of which are demonstrated in this paper.

Discriminating Spatial and Temporal Relevance in Deep Taylor Decompositions for Explainable Activity Recognition