Neural Structure
Neural structure research focuses on understanding the organization and function of neural networks, both biological and artificial, aiming to improve their design and performance. Current research emphasizes the interplay between neural structure and function, exploring how biomechanics influence neural activity (e.g., in glaucoma), how diverse neuronal populations contribute to network capabilities, and how different architectures (like U-Nets, transformers, and reservoir computing networks) impact learning and generalization across various tasks and datasets. These advancements have implications for improving diagnostic tools, developing more robust and efficient artificial intelligence, and gaining deeper insights into the workings of the brain.
Papers
Generalizability Under Sensor Failure: Tokenization + Transformers Enable More Robust Latent Spaces
Geeling Chau, Yujin An, Ahamed Raffey Iqbal, Soon-Jo Chung, Yisong Yue, Sabera Talukder
How to think step-by-step: A mechanistic understanding of chain-of-thought reasoning
Subhabrata Dutta, Joykirat Singh, Soumen Chakrabarti, Tanmoy Chakraborty