Neural networks

Neural networks are machine learning systems that perform computation across a collection of interconnected processing units, mimicking systems of neurons in a mammalian brain. The weights of the connections between processing units determine the overall behavior of the network. These weights are adaptively tuned to data through a process of "learning". The trained networks that result after learning are capable of accurately representing functional dependencies in data.

My recent work in neural networks has involved the development of network architectures that allow faster learning without a loss of predictive performance. I have also done work on visualization of neural networks, and on metrics that allow quantifying the degree of organization and symmetry in neuronal systems.

Publications on neural networks (asterisks * denote student co-authors)