Research map / Vision-language and multimodal / Human and embodied modeling
Towards robust vision by multi-task learning on monkey visual cortex
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 |
|---|---|---|---|
| Co-train classification with predicted monkey V1 responses instead of classification-only training | VGG-19 · TIN-TC | robustness score: 100% -> 118% | Helped |
| Co-train classification with real monkey V1 responses instead of classification-only training | VGG-19 · TIN-TC | robustness score: 100% -> 109% | Helped |
| Use predicted neural responses rather than real monkey V1 responses for co-training | VGG-19 multi-task learning · TIN-TC | robustness score: 109% -> 118% | Helped |
| Shuffle predicted neural responses across images (ablation) | MTL-Monkey · TIN-TC | robustness score: 118% -> 98% | Hurt |
| Co-train with Oracle-generated responses instead of classification-only training | VGG-19 · TIN-TC | robustness | Helped |
| Co-train with monkey V1 responses | MTL-Monkey · TIN-TC noise, blur, and digital corruption groups | accuracy | Helped |
| Co-train with monkey V1 responses | MTL-Monkey · TIN-TC weather corruption group | accuracy | No clear effect |
| Co-train with predicted monkey V1 responses rather than classification-only training | VGG-19 · TinyImageNet test set (clean) | accuracy: 49.26 ± 1.63% -> 46.28 ± 1.30% | Hurt |
| Co-train with Oracle-generated responses rather than classification-only training | VGG-19 · TinyImageNet test set (clean) | accuracy: 49.26 ± 1.63% -> 49.58 ± 1.39 % | No clear effect |
| Shuffle predicted neural responses across images (ablation) | MTL-Monkey · TinyImageNet test set (clean) | accuracy: 46.28 ± 1.30% -> 44.70 ± 0.61% | Hurt |
| Increase neural prediction performance by varying the amount of neural data in co-training (ablation) | MTL-Monkey · TIN-TC | robustness | Helped |
Closest papers by meaning
- Simulating a Primary Visual Cortex at the Front of CNNs Improves Robustness to Image Perturbations (NeurIPS 2020 · similarity 0.91)
- Performance-optimized deep neural networks are evolving into worse models of inferotemporal visual cortex (NeurIPS 2023 · similarity 0.91)
- Generalization in data-driven models of primary visual cortex (ICLR 2021 · similarity 0.90)
- Explicitly Modeling Subcortical Vision with a Neuro-Inspired Front-End Improves CNN Robustness (NeurIPS 2025 · similarity 0.90)
- Benchmarking Out-of-Distribution Generalization Capabilities of DNN-based Encoding Models for the Ventral Visual Cortex. (NeurIPS 2024 · similarity 0.89)
- Scaling Laws for Task-Optimized Models of the Primate Visual Ventral Stream (ICML 2025 · similarity 0.87)
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.