AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Gating Revisited: Deep Multi-layer RNNs That Can Be Trained

Mehmet Özgür Türkoglu, Stefano D’Aronco, Jan Dirk Wegner, Konrad Schindler B

TPAMI 2021 · Architectures and efficiency › RNNs and normalization · approach: Efficient recurrent architectures · 88 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 an LSTM-with-forget-gate first layer under 7 STAR layers (ablation)STAR · pMNISTaccuracy (%): 94.1 -> 95.4Improved
add an LSTM-with-forget-gate first layer under 7 STAR layers (ablation)STAR · pixel-by-pixel MNISTaccuracy (%): 99.2 -> 99.4Improved
increase STAR depth from 8 to 12 layers (ablation)STAR · pMNISTaccuracy (%): 94.1 -> 94.7Improved
increase STAR depth from 8 to 12 layers (ablation)STAR · pixel-by-pixel MNISTaccuracy (%): 99.2 -> 99.2No clear effect
increase STAR depth from 4 to 6 layers (ablation)STAR · BreizhCropsaccuracy (%): 68.2 -> 69.6Improved
increase STAR depth from 4 to 6 layers (ablation)STAR · TUM time series classificationaccuracy (%): 87.7 -> 87.6No clear effect
increase convSTAR depth from 8 to 12 layers (ablation)convSTAR · Jester gesture recognitionaccuracy (%): 92.3 -> 92.5Improved
add an LSTM first layer under 7 convSTAR layers (ablation)convSTAR · Jester gesture recognitionaccuracy (%): 92.3 -> 92.7Improved
add batch normalization between layers (ablation)IndRNN · pixel-by-pixel MNISTtraining stabilityImproved
replace LSTM with STAR cellconvLSTM (4 layers) · TUM pixel-wise classificationaccuracy (%): 90.6 -> 91.9Improved
replace IndRNN with STAR cellIndRNN (6 layers) · PennTreebank character-levelbits per character: 1.42 -> 1.30Improved
replace LSTM with STAR cell by removing output gate and coupling input and forget gates (ablation)LSTM · 12-layer MNIST gradient flowgradient magnitudeImproved

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