Friday, November 23, 11:00-12:00 in Celestijnenlaan 200A room 03.14(old number), 3001 Leuven-Heverlee
Learning Models of Activity and Functional Structure in Living Neural Networks
By Tayfun Guerel (Bernstein Center for Computational Neuroscience, Freiburg)
Cultured mammalian cortical neurons can build small closed system living neural networks. Their activity can be monitored and electrically manipulated by Multi-Electrode Arrays. In this talk, I will present machine learning algorithms that can predict the upcoming activity in both real and simulated neural networks. The core of the algorithms consists of artificial neural networks that are adapted for neuronal spike activity, i.e. time series of point events. These algorithms allow to build a hybrid system, in which artificial and biological networks interact with and adapt to each other. I will also present initial and potential examples of such an adaptive online interaction.