Research map / Optimization / Optimizers
A Theoretical Framework for Inference Learning
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 IL-SGD instead of BP-SGD | BP-SGD · CIFAR-10 batch size 1 | test accuracy (%): 36.834 -> 43.894 | Helped |
| use IL-SGD instead of BP-SGD | BP-SGD · CIFAR-10 batch size 64 | test accuracy (%): 57.044 -> 46.924 | Hurt |
| use IL-prox instead of BP-SGD | BP-SGD · CIFAR-10 batch size 1 | test accuracy (%): 36.834 -> 42.060 | Helped |
| use IL-prox + Adam instead of BP-Adam | BP-Adam · CIFAR-10 batch size 1 | test accuracy (%): 42.102 -> 42.802 | No clear effect |
| use IL-Adam instead of BP-Adam | BP-Adam · CIFAR-10 batch size 64 | test accuracy (%): 56.466 -> 54.066 | Hurt |
| use IL-prox + Adam instead of BP-Adam | BP-Adam · CIFAR-10 batch size 64 | test accuracy (%): 56.466 -> 55.572 | Hurt |
| use IL-prox instead of IL-SGD | IL-SGD · CIFAR-10 batch size 64 | test accuracy (%): 46.924 -> 49.844 | Helped |
| use IL-prox instead of BP-SGD | BP-SGD · CIFAR-10 autoencoder | reconstruction BCE: 1937.884 -> 1878.463 | Helped |
| use IL-SGD instead of BP-SGD | BP-SGD · CIFAR-10 autoencoder | reconstruction BCE: 1937.884 -> 2004.062 | Hurt |
| use IL-prox fast instead of BP-Adam | BP-Adam · CIFAR-10 autoencoder | reconstruction BCE: 1912.202 -> 1875.899 | Helped |
| increase learning rate from .01 to 10 for IL-prox | IL-prox · CIFAR-10 stability | test accuracy (%): 34.97 -> 37.01 | No clear effect |
| increase learning rate from .01 to 10 for BP-prox | BP-prox · CIFAR-10 stability | test accuracy (%): 38.39 -> 12.21 | Hurt |
Closest papers by meaning
- Understanding and Improving Optimization in Predictive Coding Networks (AAAI 2024 · similarity 0.92)
- Brain-like Variational Inference (NeurIPS 2025 · similarity 0.87)
- Forward Learning with Top-Down Feedback: Empirical and Analytical Characterization (ICLR 2024 · similarity 0.86)
- Reducing Fine-Tuning Memory Overhead by Approximate and Memory-Sharing Backpropagation (ICML 2024 · similarity 0.86)
- Towards Exact Gradient-based Training on Analog In-memory Computing (NeurIPS 2024 · similarity 0.85)
- Boosting Inference with Guided Reasoning: Stochastic Exploration for Recursive Models (NeurIPS 2026 · 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.