AI Paper Atlas

Research map / Architectures and efficiency / Neural architecture search

Searching for Two-Stream Models in Multivariate Space for Video Recognition

Xinyu Gong, Heng Wang, Mike Zheng Shou, Matt Feiszli, Zhangyang Wang, Zhicheng Yan

ICCV 2021 · Architectures and efficiency › Neural architecture search · approach: Bi-level differentiable NAS · 10 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
add a fast stream and searched fusion blocks to the slow-stream model (ablation)Auto-TSNet · Kinetics-400top-1 accuracy (%): 70.8 -> 74.1 (+3.3)Improved
add searched GloRe attention blocks (ablation)Auto-TSNet step-2 model · Kinetics-400top-1 accuracy (%): 74.1 -> 74.6 (+0.5)Improved
search the backbone instead of hand-crafting it (ablation)Slow stream · Kinetics-400top-1 accuracy (%): 70.2 -> 70.8Improved
search fusion locations instead of manual placement (ablation)Manual-TSNet-70% · Kinetics-400top-1 accuracy (%): 73.2 -> 73.7Improved
search GloRe locations instead of no attention (ablation)Manual-TSNet-70% · Kinetics-400top-1 accuracy (%): 73.2 -> 73.6Improved
use progressive search instead of one-step search (ablation)Auto-TSNet-S · Kinetics-400top-1 accuracy (%): 72.8 -> 75.4Improved
use progressive search instead of one-step search (search time) (ablation)Auto-TSNet-S · Kinetics-400search days: 6.5 -> 4.8Improved
replace single-stream X3D-S with a manual two-stream model (70% FLOPS to sparse stream)Manual-TSNet · Kinetics-400top-1 accuracy (%): 72.9 -> 73.2Improved
lower the sparse-stream FLOPS ratio from 70% to 55%Manual-TSNet · Kinetics-400top-1 accuracy (%): 73.2 -> 72.4Hurt
increase manual fusion blocks from 5 to 10 (ablation)Manual-TSNet-70% · Kinetics-400top-1 accuracy (%): 73.2 -> 71.2Hurt
add 5 manual fusion blocks to a model with none (ablation)Manual-TSNet-70% · Kinetics-400top-1 accuracy (%): 70.7 -> 73.2Improved
manually add two GloRe blocks in stage 2 only (ablation)Manual-TSNet-70% · Kinetics-400top-1 accuracy (%): 73.2 -> 72.7Hurt

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