AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Learning Hierarchical Information Flow with Recurrent Neural Modules

Danijar Hafner, Alexander Irpan, James Davidson, Nicolas Heess

NeurIPS 2017 · Architectures and efficiency › RNNs and normalization · approach: Brain-inspired recurrent architectures · 4 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
Use ThalNet instead of a standard GRU baselineLanguage modeling · text8BPC: 1.41 -> 1.39Improved
Use ThalNet instead of a stacked GRU baselineSequential image classification · Sequential Permuted MNISTtest accuracyImproved
Use ThalNet instead of a stacked GRU baselineSequential image classification · Sequential CIFAR-10test accuracyImproved
Use recurrent modules instead of fully connected-only modules (ablation)ThalNet · Sequential MNISTtraining speed and stabilityImproved
Add a fully connected layer before the GRU module (ablation)ThalNet · Sequential MNISTtest accuracyImproved
Add a fully connected layer after the GRU module (ablation)ThalNet · Sequential MNISTtest accuracyNo clear effect
Use weight-normalized reading instead of linear or fast-softmax reading (ablation)ThalNet · Sequential MNISTtraining stabilityImproved
Use fast Gaussian reading instead of the other tested reading mechanisms (ablation)ThalNet · Sequential MNISTtest accuracyHurt

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