AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Biologically Motivated Algorithms for Propagating Local Target Representations

Alexander G. Ororbia, Ankur Mali

AAAI 2019 · Architectures and efficiency › RNNs and normalization · approach: Brain-inspired network comparison · 61 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 error-driven local target alignment (LRA-E) instead of the compared backprop alternatives3-, 5-, and 8-layer MLPs · MNISTtest error: 3.02 -> 2.20; 2.98 -> 1.97; 3.76 -> 2.55Improved
Use error-driven local target alignment (LRA-E) instead of the compared backprop alternatives5- and 8-layer MLPs · Fashion MNISTtest error: 12.89 -> 12.31; 13.01 -> 12.69Improved
Use LRA-E instead of the strongest listed baseline3-layer MLP · Fashion MNISTtest error: 12.74 -> 13.51Hurt
Replace fixed DTP corruption noise with adaptive layerwise noise (DTP-σ) (ablation)3-, 5-, and 8-layer MLPs · MNISTtest error: 2.8 -> 2.38; 4.94 -> 2.57; 10.1 -> 2.56Improved
Replace fixed DTP corruption noise with adaptive layerwise noise (DTP-σ) (ablation)3-, 5-, and 8-layer MLPs · Fashion MNISTtest error: 15.03 -> 13.95; 19.66 -> 12.99; 17.58 -> 13.01Mixed
Use Adam instead of SGD (ablation)LRA-E · MNIST, 3 layerstest error: 2.20 -> 1.75Improved
Use RMSprop instead of SGD (ablation)LRA-E · MNIST, 3 layerstest error: 2.20 -> 2.02Improved
Use Adam instead of SGD (ablation)LRA-E · Fashion MNIST, 3 layerstest error: 13.51 -> 12.42Improved
Use RMSprop instead of SGD (ablation)LRA-E · Fashion MNIST, 3 layerstest error: 13.51 -> 14.90Hurt
Use Cauchy rather than Gaussian hidden-layer local losses (ablation)LRA-E · Classification experimentsrepresentation sparsityImproved
Use DTP-σ rather than DTP across different weight initializations (ablation)5- and 8-layer MLPs · MNIST and Fashion MNISTinitialization sensitivityImproved

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