Neural network models of the thalamocortical circuits.
Brain-inspired algorithms for continual learning, generalization and task discovery. "Thalamus" is an algorithm inspired from the thalamocortical circuit that solves an important machine learning problem. How to identify tasks in a stream of unlabelled data and jointly learn the most practical clustering of tasks based on similarity while simultanuously learning the tasks.
Analogical reasoning in neural networks. A joint project with Michelanglo.
Context switching in Schizophrenia patients. In a collaboration with Mathew Nassar.