AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Scaling Direct Feedback Learning with Theoretical Guarantees

Paul Caillon, Erwan Fagnou, Blaise Delattre, Alexandre Allauzen

ICLR 2026 · Architectures and efficiency › RNNs and normalization · approach: Backprop-free credit assignment

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 BP calibration (T=1) to GrAPE (ablation)GrAPE on AlexNet · CIFAR-100accuracy: 45.45 -> 62.63Improved
add BP calibration (T=1) to GrAPE (ablation)GrAPE on VGG-16 · CIFAR-100accuracy: 32.40 -> 56.93Improved
add BP calibration (T=1) to DFA (ablation)DFA on VGG-16 · CIFAR-100accuracy: 1.00 -> 29.40Improved
add BP calibration (T=1) to DFA (ablation)DFA on AlexNet · CIFAR-100accuracy: 42.59 -> 49.37Improved
replace DFA fixed feedback with GrAPE learned aligned feedbackDFA on AlexNet · CIFAR-100accuracy: 42.59 -> 45.45Improved
replace DFA fixed feedback with GrAPE learned aligned feedbackDFA with T=1 calibration on VGG-16 · CIFAR-100accuracy: 29.40 -> 56.93Improved
use GrAPE instead of DFA, no calibrationDFA on ResNet-20 · CIFAR-100accuracy: 20.94 -> 24.28Improved
add BP calibration (T=1) to GrAPE (ablation)GrAPE on ResNet-20 · CIFAR-100accuracy: 24.28 -> 64.82Improved
increase calibration interval from T=1 to T=50 (ablation)GrAPE + calibration on ResNet-20 · CIFAR-100accuracy: 64.82 -> 36.79Hurt
add BP calibration (T=1) to GrAPE (ablation)GrAPE on Transformer-Base macro · WikiText-103validation perplexity: 42.3 -> 33.1Improved
add BP calibration (T=1) to GrAPE (ablation)GrAPE on Transformer-Base micro · WikiText-103validation perplexity: 81.1 -> 67.3Improved
increase calibration interval from T=1 to T=10 (ablation)GrAPE + calibration on Transformer-Base micro · WikiText-103validation perplexity: 67.3 -> 78.2Hurt

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