Research map / Probabilistic and scientific modeling / Differential equations and PDEs
Focus of Attention Improves Information Transfer in Visual Features
Matteo Tiezzi, Stefano Melacci, Alessandro Betti, Marco Maggini, Marco Gori
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 |
|---|---|---|---|
| train with FOA density instead of uniform | architecture D · SparseMNIST (test FOA) | MI index: 0.103 -> 0.431 | Improved |
| train with FOA density instead of uniform | architecture S · SparseMNIST (test FOA) | MI index: 0.239 -> 0.486 | Improved |
| train with FOA density instead of uniform | architecture D · Call (test FOA) | MI index: 0.430 -> 0.582 | Improved |
| train with FOA density instead of uniform | architecture S · Call (test FOA) | MI index: 0.405 -> 0.405 | No clear effect |
| train with FOA density instead of uniform | architecture S · Carpark (test FOA) | MI index: 0.742 -> 0.675 | Hurt |
| train with FOA density instead of uniform | architecture D · Carpark (test FOA) | MI index: 0.678 -> 0.639 | Hurt |
| train with FOAW (window) density instead of uniform | architecture D · Carpark (test FOA) | MI index: 0.678 -> 0.768 | Improved |
| train with FOA density instead of uniform | architecture DL · Call (test FOA) | MI index: 0.246 -> 0.365 | Improved |
| use AVG instead of PLA temporal locality (ablation) | architecture D, FOA · SparseMNIST (test FOA) | MI index: 0.149 -> 0.321 | Improved |
| use VAR instead of AVG temporal locality (ablation) | architecture D, FOA · SparseMNIST (test FOA) | MI index: 0.321 -> 0.431 | Improved |
| use AVG instead of PLA temporal locality (ablation) | architecture D, FOA · Carpark (test FOA) | MI index: 0.458 -> 0.639 | Improved |
| use PLA instead of AVG temporal locality (ablation) | architecture DL, FOA · Call (test FOA) | MI index: 0.214 -> 0.365 | Improved |
Closest papers by meaning
- Flexible Context-Driven Sensory Processing in Dynamical Vision Models (NeurIPS 2024 · similarity 0.85)
- Learning 1D Causal Visual Representation with De-focus Attention Networks (NeurIPS 2024 · similarity 0.85)
- Degrees of Freedom for Linear Attention: Distilling Softmax Attention with Optimal Feature Efficiency (NeurIPS 2025 · similarity 0.84)
- Bayes optimal learning of attention-indexed models (NeurIPS 2025 · similarity 0.84)
- On the Surprising Effectiveness of Attention Transfer for Vision Transformers (NeurIPS 2024 · similarity 0.84)
- High-Dimensional Learning Dynamics of Attention-Indexed Models (NeurIPS 2026 · similarity 0.84)
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.
