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

Training competitive deep video models is an order of magnitude slower thantraining their counterpart image models. Slow training causes long researchcycles, which hinders progress in video understanding research. Followingstandard practice for training image models, video model training assumes afixed mini-batch shape: a specific number of clips, frames, and spatial size.However, what is the optimal shape? High resolution models perform well, buttrain slowly. Low resolution models train faster, but they are inaccurate.Inspired by multigrid methods in numerical optimization, we propose to usevariable mini-batch shapes with different spatial-temporal resolutions that arevaried according to a schedule. The different shapes arise from resampling thetraining data on multiple sampling grids. Training is accelerated by scaling upthe mini-batch size and learning rate when shrinking the other dimensions. Weempirically demonstrate a general and robust grid schedule that yields asignificant out-of-the-box training speedup without a loss in accuracy fordifferent models (I3D, non-local, SlowFast), datasets (Kinetics,Something-Something, Charades), and training settings (with and withoutpre-training, 128 GPUs or 1 GPU). As an illustrative example, the proposedmultigrid method trains a ResNet-50 SlowFast network 4.5x faster (wall-clocktime, same hardware) while also improving accuracy (+0.8% absolute) onKinetics-400 compared to the baseline training method. Code is availableonline.

A Multigrid Method for Efficiently Training Video Models