Research map / Vision-language and multimodal / Human and embodied modeling
A Neurobiological Evaluation Metric for Neural Network Model Search
Check 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 HMS to rank hyperparameterized PredNets | PredNet · KITTI next-frame prediction and Gazoobian object matching | Pixel MSE and object matching accuracy: Pixel MSE: 0.092 (0.148) -> 0.009 (0.003); object matching accuracy: 0.367 (0.134) -> 0.459 (0.049) | Helped |
| Increase HMS across sampled networks | PredNet · 95 randomly hyperparameterized PredNets | Pixel MSE: -0.646** | Helped |
| Increase HMS across sampled networks | PredNet · 95 randomly hyperparameterized PredNets | Object matching accuracy: 0.575** | Helped |
| Select networks with the lowest HMS | PredNet · Ten lowest-HMS models | Pixel MSE and object matching accuracy: Pixel MSE: 0.314 ($\mathrm{SD} = 0.138$); object matching accuracy: 0.13 ($0.15$) | Hurt |
| Use HMS as an early-stopping signal | PredNet training · 95 PredNets, post hoc analysis | Training time: 67% | Helped |
| Use the HMS threshold of 0.161 to screen models | PredNet · 95 randomly hyperparameterized PredNets | Object matching accuracy and pixel MSE: 0.161; 13 of the 95 models (13.7%) | Mixed |
| Control for learning rate in the HMS-performance relationship | PredNet · 1811-model partial-correlation sample | Correlation of HMS with computer vision performance: $p < 0.001$ | No clear effect |
| Evaluate HMS early during training rather than waiting for task metrics to stabilize | PredNet training · 74 four-layer PredNets | Epochs until metric stabilizes: HMS: 33 epochs; accuracy: 66.5 epochs ($\mathrm{SD} = 36$) | Helped |
Closest papers by meaning
- Scaling Laws for Task-Optimized Models of the Primate Visual Ventral Stream (ICML 2025 · similarity 0.87)
- Characterizing Universal Object Representations Across Vision Models (NeurIPS 2026 · similarity 0.87)
- Not all solutions are created equal: An analytical dissociation of functional and representational similarity in deep linear neural networks (ICML 2025 · similarity 0.87)
- Differentiable Optimization of Similarity Scores Between Models and Brains (ICLR 2025 · similarity 0.87)
- Model–Behavior Alignment under Flexible Evaluation: When the Best-Fitting Model Isn’t the Right One (NeurIPS 2025 · similarity 0.87)
- A Convolutional Neural Network Interpretable Framework for Human Ventral Visual Pathway Representation (AAAI 2024 · similarity 0.86)
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.