Overview of the lab of Uludag & Spiking Neuronal Network Modeling (CANCELLED)
In the first part of this talk, Dr. Kamil Uludag will present an overview of his group. Specifically, he will present his work on analysis methods for human brain effective connectivity from functional MRI data. To that end, the published physiological dynamic causal modelling (DCM) approach will be introduced. DCM is inspired by experimental observations about the physiological underpinnings of the fMRI signal to study effective connectivity in the brain. Ongoing projects overcoming the limitations of existing DCM approaches will be discussed. In the second part of the talk, Dr. Soheila Nazari (postdoc in Uludag’s lab) will present a novel un-supervised Spiking Neural Network, which can be utilized for pattern recognition, development of novel problem-solving techniques and classification and can be applied on neuronal or fMRI data. In contrast to deep learning approaches, the spiking neural network (SNN) consists of more biologically plausible units, such as excitatory and inhibitory synapses based on AMPA and GABA synaptic currents. The spiking network architecture contains image coding and network layers and utilizes un-supervised spatial-temporal learning mechanisms. This novel network is then applied to classify MNISTdata and compared with existing approaches.