AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

STCN: Stochastic Temporal Convolutional Networks

Emre Aksan, Otmar Hilliges

ICLR 2019 · Architectures and efficiency › RNNs and normalization · approach: Efficient recurrent architectures · 16 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
connect all hierarchical latent samples directly to the output instead of using only the lowest-level sample (ablation)STCN · IAM-OnDBELBO: ≥ 1339 -> ≥ 1796Improved
connect all hierarchical latent samples directly to the output instead of using only the lowest-level sample (ablation)STCN · Deepwritingaverage log-likelihood per sequence: ≥ 605 -> ≥ 797Improved
connect all hierarchical latent samples directly to the output instead of using only the lowest-level sample (ablation)STCN · TIMITELBO: ≥ 69195 -> ≥ 71386Improved
connect all hierarchical latent samples directly to the output instead of using only the lowest-level sample (ablation)STCN · Blizzardaverage log-likelihood per sequence: 15800 -> 16288Improved
replace TCN blocks with LSTM cells while retaining the latent hierarchy and dense connections (ablation)STCN-dense · TIMITELBO: ≥ 71386 -> ≥ 50113Hurt
replace TCN blocks with LSTM cells while retaining the latent hierarchy and dense connections (ablation)STCN-dense · IAM-OnDBELBO: ≥ 1796 -> ≥ 1581Hurt
replace TCN blocks with LSTM cells in the 25x256 configuration (ablation)STCN-dense · TIMITELBO: ≥ 71386 -> ≥ 26240Hurt
replace TCN blocks with LSTM cells in the 25x256 configuration (ablation)STCN-dense · IAM-OnDBELBO: ≥ 1796 -> ≥ 1377Hurt
replace concatenation of latent samples with summation before the output (ablation)STCN-dense · sequence modeling experimentsELBOHurt
increase speech-model convolution filters from 256 to 512 (ablation)STCN-dense · TIMITaverage log-likelihood per sequence: ≥ 71386 -> ≥ 77438Improved
increase speech-model convolution filters from 256 to 512 (ablation)STCN-dense · Blizzardaverage log-likelihood per sequence: ≥ 16288 -> ≥ 17670Improved

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