AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Scalable Real-Time Recurrent Learning Using Columnar-Constructive Networks

Khurram Javed, Haseeb Shah, Richard S. Sutton, Martha White

JMLR 2023 · 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
combine columnar and constructive networks into CCNsRTRL LSTM learners · Trace patterningprediction errorImproved
CCNs instead of T-BPTTLSTM state construction · Atari prediction 16x16normalized return error: over 5x lower error in many environments; Breakout error 20% moreImproved
add online feature normalization to CCNs (ablation)CCN · Trace patterningprediction errorImproved
add online feature normalization to CCNs (ablation)CCN · Atari prediction 16x16 (10M steps)prediction errorImproved
increase T-BPTT truncation length with fixed 8 features (ablation)T-BPTT · Atari prediction 16x16prediction error: truncation 2 to 15Improved
increase T-BPTT feature count with truncation 8 (ablation)T-BPTT · Atari prediction 16x16prediction error: 2 features error twice that of 15 featuresImproved
large networks with short truncation (3:10, 5:8) vs small networks with long truncation (15:4) (ablation)T-BPTT · Trace patterningprediction error: 3:10 and 5:8 perform worst; 15:4, 30:2, 20:3 betterHurt
allow T-BPTT 10 features with longer truncation (more compute) (ablation)T-BPTT · Trace patterningprediction error: nearly matches CCN at around seven times more computeImproved
increase features-per-stage from 1 to 6 (ablation)CCN · Trace patterningprediction error: 1 to 6 featuresImproved
reduce steps-per-stage to 0.5 million (ablation)CCN (2 features-per-stage) · Trace patterningasymptotic error: 0.5 million steps-per-stageMixed
add hierarchical features (CCN) over one-feature-at-a-time Constructive networks (ablation)Constructive networks · Trace patterningprediction errorImproved

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