Matching the Neuronal Representations of V1 is Necessary to Improve Robustness in CNNs with V1-like Front-ends
See where this sits in the topic map →Summary AI-generated
- TL;DR
- While convolutional neural networks excel at object recognition, they struggle when identifying objects in images corrupted by common noise patterns.
- Problem
- Standard convolutional neural networks struggle to identify objects in images corrupted by common noise patterns.
- Method
- The authors built two variants of a model featuring a front-end that models the primate primary visual cortex (V1): one samples receptive field properties uniformly, while the other samples from empirical biological distributions.
- Results
- The model using biological sampling showed an 8.72% higher relative robustness to image corruptions compared to the uniform variant.
- Contributions
- The study demonstrates that precisely matching the distribution of receptive field properties found in primate V1 is key to improving model robustness.
- Limitations
- Not specified in the abstract.
- Takeaways
- This finding clarifies the origin of robustness improvements in biologically inspired models, highlighting the need to precisely mimic neuronal representations found in the primate brain.
- Applications
- Not specified in the abstract.
- Topics
- Convolutional Neural Networks; Primary Visual Cortex; Model Robustness; Neuronal Representations
- For industry
- Not specified in the abstract.
- Why it matters
- Not specified in the abstract.
Abstract
While some convolutional neural networks (CNNs) have achieved great success in object recognition, they struggle to identify objects in images corrupted with different types of common noise patterns. Recently, it was shown that simulating computations in early visual areas at the front of CNNs leads to improvements in robustness to image corruptions. Here, we further explore this result and show that the neuronal representations that emerge from precisely matching the distribution of RF properties found in primate V1 is key for this improvement in robustness. We built two variants of a model with a front-end modeling the primate primary visual cortex (V1): one sampling RF properties uniformly and the other sampling from empirical biological distributions. The model with the biological sampling has a considerably higher robustness to image corruptions that the uniform variant (relative difference of 8.72%). While similar neuronal sub-populations across the two variants have similar response properties and learn similar downstream weights, the impact on downstream processing is strikingly different. This result sheds light on the origin of the improvements in robustness observed in some biologically-inspired models, pointing to the need of precisely mimicking the neuronal representations found in the primate brain.