Research
MIT CBMM
Computational Models of Visual Perception
Overview
As a research intern at the MIT Center for Brains, Minds, and Machines (CBMM), I evaluated deep learning models for visual scene classification and explored how stylistic transformations influence learned representations.
I proposed a style-transfer-based approach to probe model sensitivity to texture versus structure, connecting machine learning performance with perceptual principles inspired by neuroscience. This work bridged computational modeling with cognitive science.
Key Contributions
- Evaluated 15+ deep learning architectures for scene classification
- Studied representational robustness under style transformations
- Connected computational models with perceptual and cognitive insights
- Conducted literature review and experimental validation
Technologies
Python
PyTorch
CNNs
Style Transfer
Computer Vision