Tutorial

	Dr. Giacomo Indiveri Institute of Neuroinformatics University of Zurich / ETH
Title	Analog VLSI circuits for spiking neural networks
Abstract	This tutorial will cover the design of analog circuits for implementing networks of spiking neurons in VLSI technology. It will begin with an overview of past and recent analog VLSI implementations of neuron models, then describe potential applications for systems comprising multi-neuron VLSI chips, and then focus on analog circuit solutions, using transistors operated in the weak-inversion regime. This part of the tutorial will be subdivided in four sub-parts: we will first outline the basic subthreshold characteristics of MOSFETs, pointing out the analogies between the biophysics of biological neurons and the physics of transistor channels. In the second part we will describe basic subthreshold circuits, log-domain filters and non-linear temporal filters for implementing synaptic dynamics. In the third part we will explain the operation of hybrid analog-digital circuits which implement Integrate-and-Fire neuron models. We will conclude the tutorial in the fourth part, with the description of exponential I&F circuits, spike-frequency adaptation and refractory period mechanisms, and circuits for implementing spike-based learning mechanisms.
Bio	Giacomo Indiveri is a Senior Lecturer at the Institute of Neuroinformatics of the University of Zurich and ETH Zurich. He graduated in electrical engineering from the University of Genoa, Italy in 1992 and worked as a research fellow in the division of Biology with Prof. Christof Koch, at the California Institute of Technology from 1994 to 1996. In 1996 he moved to the Institute of Neuroinformatics, University of Zurich and ETH Zurich. He obtained a PhD in Electrical Engineering and Computer Science in 2004 from the University of Genova, Italy and the Habilitation and "venia legendi" in Neuromorphic Engineering in 2006 from ETH Zurich. Indiveri has been working on the design of biologically inspired neuromorphic systems, neuromorphic cognitive architectures, and artificial neural VLSI systems for over 15 years.