Research map / Architectures and efficiency / RNNs and normalization
Biologically Motivated Algorithms for Propagating Local Target Representations
Alexander G. Ororbia, Ankur Mali
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 changed | Where | Measured | Effect |
|---|---|---|---|
| Use error-driven local target alignment (LRA-E) instead of the compared backprop alternatives | 3-, 5-, and 8-layer MLPs · MNIST | test error: 3.02 -> 2.20; 2.98 -> 1.97; 3.76 -> 2.55 | Improved |
| Use error-driven local target alignment (LRA-E) instead of the compared backprop alternatives | 5- and 8-layer MLPs · Fashion MNIST | test error: 12.89 -> 12.31; 13.01 -> 12.69 | Improved |
| Use LRA-E instead of the strongest listed baseline | 3-layer MLP · Fashion MNIST | test error: 12.74 -> 13.51 | Hurt |
| Replace fixed DTP corruption noise with adaptive layerwise noise (DTP-σ) (ablation) | 3-, 5-, and 8-layer MLPs · MNIST | test error: 2.8 -> 2.38; 4.94 -> 2.57; 10.1 -> 2.56 | Improved |
| Replace fixed DTP corruption noise with adaptive layerwise noise (DTP-σ) (ablation) | 3-, 5-, and 8-layer MLPs · Fashion MNIST | test error: 15.03 -> 13.95; 19.66 -> 12.99; 17.58 -> 13.01 | Mixed |
| Use Adam instead of SGD (ablation) | LRA-E · MNIST, 3 layers | test error: 2.20 -> 1.75 | Improved |
| Use RMSprop instead of SGD (ablation) | LRA-E · MNIST, 3 layers | test error: 2.20 -> 2.02 | Improved |
| Use Adam instead of SGD (ablation) | LRA-E · Fashion MNIST, 3 layers | test error: 13.51 -> 12.42 | Improved |
| Use RMSprop instead of SGD (ablation) | LRA-E · Fashion MNIST, 3 layers | test error: 13.51 -> 14.90 | Hurt |
| Use Cauchy rather than Gaussian hidden-layer local losses (ablation) | LRA-E · Classification experiments | representation sparsity | Improved |
| Use DTP-σ rather than DTP across different weight initializations (ablation) | 5- and 8-layer MLPs · MNIST and Fashion MNIST | initialization sensitivity | Improved |
Closest papers by meaning
- Efficient Target Propagation by Deriving Analytical Solution (AAAI 2024 · similarity 0.88)
- Understanding Sample Efficiency in Predictive Coding (NeurIPS 2026 · similarity 0.86)
- Implicit Regularization in Feedback Alignment Learning Mechanisms for Neural Networks (ICML 2024 · similarity 0.86)
- Forward Learning with Top-Down Feedback: Empirical and Analytical Characterization (ICLR 2024 · similarity 0.86)
- Improving equilibrium propagation without weight symmetry through Jacobian homeostasis (ICLR 2024 · similarity 0.85)
- Dendritic Localized Learning: Toward Biologically Plausible Algorithm (ICML 2025 · similarity 0.85)
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.
