Research
Continual Learning
Advance biologically and cognitively inspired approaches to lifelong learning, developing memory-augmented architectures, meta-learning methods, and benchmark suites to enable continual adaptation while mitigating catastrophic forgetting
Foundation Models for Neuroscience
Advance scalable neurofoundation models that integrate neural activity, behavior, and environmental data through multimodal representation learning, large-scale data integration, and standardized evaluation to enable generalizable, task-flexible models of brain function
Language & Vision
Advance integrating perception, language, and interaction to develop human-centered AI, supporting four projects spanning neural language processing, multimodal interaction, safe and interpretable dialogue, robust perception, and assistive technologies, while fostering new interdisciplinary collaborations around shared multimodal challenges
Neural Mechanisms of Intelligence
Investigates the computational principles underlying biological intelligence through projects on connectome-informed architectures, circuit dynamics, modular learning, and embodied control, uncovering how neural structure and dynamics enable efficient, flexible, and generalizable computation
