Neuromorphic Circuit Design

Faculteit Science and Engineering
Jaar 2022/23
Vakcode WMPH044-05
Vaknaam Neuromorphic Circuit Design
Niveau(s) master
Voertaal Engels
Periode semester II b

Uitgebreide vaknaam Neuromorphic Circuit Design
Leerdoelen At the end of the course, the student is able to:
1. discuss the general aims of neuromorphic systems, with particular focus on full-custom silicon implementations of spiking neural networks;
2. recognize advantages and drawbacks of CMOS neuromorphic systems and compare them with other approaches;
3. draw the neuromorphic circuits treated in the course;
4. use the MOS sub-threshold equation to derive the transfer function of the neuromorphic circuits treated in the course;
5. discuss the behavior of the neuromorphic circuits treated in the course, identify their main characteristics and how they are affected by circuit parameters;
6. execute SPICE simulations using Cadence: analyze simulation results and discuss deviations from theory.
Omschrijving Today's most powerful computers are still outperformed by biological brains in routine functions such as vision, audition, and motor control. Understanding the principles of biological computation and how to implement them in hardware, is crucial for comprehending biology but also for developing novel techniques for information processing. Neuromorphic engineering attempts to use the principles of computation observed in biology. It emphasizes distributed, collective, self-organized, event-driven mechanisms. The building blocks of neuromorphic systems are analog circuits in which transistors are mostly operated in weak inversion (below threshold), where their exponential I-V characteristics and low currents can be exploited. These features allow the implementation of massively parallel, low-power spiking recurrent neural networks as well as efficient sensors. This course is an introduction to the design of sub-threshold circuits for the physical emulation of neural computation.

The course presents a broad view of various circuits used in neuromorphic engineering and their applications. The course comprises the following parts:

Part 1 - Introduction and definition of the classic neuromorphic design and MOS weak inversion/subthreshold physics;

Part 2 - Circuit building blocks used in weak inversion regime (current-source, current mirror, source follower, differential pair, transconductance amplifiers);

Part 3 - Neuromorphic building blocks used for neural emulation (differential pair integrator (DPI), DPI synapse, axon-hillock neuron, integrate-and-fire neuron, low-power neuron, DPI neuron).
Uren per week
Onderwijsvorm Hoorcollege (LC), Opdracht (ASM), Practisch werk (PRC), Werkcollege (T)
(16 LC, 8 T, 20 ASM, 24 PRC, 72 self-study)
Toetsvorm Mondeling tentamen (OR), Opdracht (AST), Practisch werk (PR)
(The practical is mandatory. Arrangements should be taken to take a missed practical.)
Vaksoort master
Coördinator Prof. Dr. E. Chicca
Docent(en) Prof. Dr. E. Chicca
Verplichte literatuur
Titel Auteur ISBN Prijs
Event-based neuromorphic systems (Wiley; 2015) Shih-Chii Liu, Tobi Delbruck, Giacomo Indiveri, Adrian Whatley, Rodney Douglas 978-0-470-01849-1
Lecture slides made available by the lecturer.
Article: Synaptic dynamics in analog VLSI. Neural computation, 19(10), 2581-2603 (2007) C. Bartolozzi and G. Indiveri
Article: Neuromorphic electronic circuits for building autonomous cognitive systems. Proceedings of the IEEE, 102(9), 1367-1388. (2014) E. Chicca, F. Stefanini, C. Bartolozzi and G. Indiveri
Analog VLSI: circuits and principles (Mit Press Ltd; 2002) Shih-Chii Liu, Jörg Kramer, Giacomo Indiveri, Tobias Delbruck, and Rodney Douglas 9780262122559
Entreevoorwaarden Bachelor in Physics, Bachelor in Applied Physics, Bachelor in Chemistry, Bachelor in Chemical Engineering, Bachelor in Industrial Engineering and Management, Bachelor of Computing Science, Bachelor in Artificial Intelligence, Bachelor in Information Science or equivalent.
Opgenomen in
Opleiding Jaar Periode Type
MSc Applied Physics  ( Keuzevakken in Bio-inspired Design for Future Technologies) - semester II b keuze: BDFT
MSc Artificial Intelligence  (C - Elective Course Units) - semester II b keuze
MSc Nanoscience  (Optional Courses) - semester II b keuze