Alara Kaymak

Data Science Graduate Student

Scroll for more info

About

I am a Data Science graduate student at Vanderbilt University with a focus on image data and spatial data analysis. My work spans computer vision, vision transformers, and geospatial modeling, as well as multimodal learning across language and signal data. I am interested in how neuroscience can inspire more efficient approaches to machine learning, especially in projects that connect human brain activity with image and spatial representations.

Research Interests

Neural Network Architectures Brain Alignment Multimodal Integration Cognitive Science Machine Learning Computer Vision AR/VR

Current Research Projects

Personalized Representational Connectivity with fMRI-Guided CLIP

Human visual perception emerges from coordinated activity across many interconnected brain regions. To develop AI models that better reflect this biological organization, we fine-tuned CLIP to predict both the representational structure within individual visual cortical areas and the functional connectivity patterns linking them. Using lightweight model adaptations—such as layer-specific feature reweighting and low-rank personalization—the model successfully captured individual differences in neural processing across 14 visual regions. Additionally, the fMRI-aligned models achieved zero-shot generalization to MEG dynamics, demonstrating a pathway toward personalized and biologically grounded multimodal AI systems.