AI Paper Atlas

Research map / Video / Action recognition

Long Movie Clip Classification with State-Space Video Models

Md Mohaiminul Islam, Gedas Bertasius

ECCV 2022 · Video › Action recognition · approach: Contrastive video representation learning · 109 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace self-attention with S4 layers in the long-range decoderLong Sequence Transformer · LVU Relationtop-1 accuracy: 52.38 -> 57.14Improved
replace self-attention with S4 layers in the long-range decoderLong Sequence Transformer · LVU Scenetop-1 accuracy: 62.79 -> 67.44Improved
replace self-attention with S4 layers in the long-range decoderLong Sequence Transformer · LVU Directortop-1 accuracy: 56.07 -> 62.61Improved
replace self-attention with S4 layers in the long-range decoderLong Sequence Transformer · LVU Like ratioMSE: 0.31 -> 0.26Improved
use S4 instead of self-attention at 11,760 tokensLong Sequence Transformer · LVU efficiencyGPU memory (GB): 41.38 -> 5.15Improved
use S4 instead of self-attention at 11,760 tokensLong Sequence Transformer · LVU efficiencytraining samples/s: 1.88 -> 4.95Improved
add multi-scale pooling and channel scaling to vanilla S4 decoder (ablation)ViS4mer · LVU content taskstop-1 accuracy: 49.53 -> 55.12Improved
add multi-scale pooling and channel scaling to vanilla S4 decoder (ablation)ViS4mer · LVU user engagementMSE: 2.30 -> 1.94Improved
add pooling alone, without channel scaling (ablation)vanilla S4 · LVU content taskstop-1 accuracy: 49.53 -> 48.96Hurt
add channel scaling alone, without pooling (ablation)vanilla S4 · LVU content taskstop-1 accuracy: 49.53 -> 52.25Improved
replace S4 with Performer (ablation)ViS4mer · LVU Relationtop-1 accuracy: 57.14 -> 50.00Hurt
replace S4 with Orthoformer (ablation)ViS4mer · LVU Genretop-1 accuracy: 54.71 -> 55.79No clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy