Neuromorphic Engineering: From Neural Systems to Brain-Like Engineered Systems

The styles of computation used by biological systems are fundamentally different from those used by conventional computers: biological neural networks process information using energyefficient, highly parallel, event-driven architectures as opposed to clocked serial processing. They are composed ofmultiple instances of heterogeneous elements and are able to self-repair, adapt and learn from the interaction with the environment. Memory in biological systems is distributed throughout the architecture, relying on local bio-chemicalmachinery for efficient storage and recall. These remarkable biological traits yield a potentially attractive alternative to conventional computing strategies. A special focus of this issue is Neuromorphic VLSI systems that are composed of Very Large Scale Integrated (VLSI) devices with hybrid analog/digital circuits that implement hardware models of biological systems. When implemented in VLSI (including FPGA) technology, neuromorphic systems often exploit strategies similar to those observed in biological systems for maximizing compactness, optimizing robustness to noise, minimizing power consumption, and increasing fault tolerance (Mead, 1990). By emulating the neural style of computation, neuromorphic architectures can exploit to the fullest potential the features of advanced scaled VLSI processes and future emerging technologies, naturally copingwith the problems that characterize them, such as time dependent device variability, and mismatch. In this Special Issuewe expressly called for a broad range of topics related to the common theme of Neuromorphic Engineering. The various contributions received actually describe recent developments and progress in understanding the interplay between biology and technology for the developments of bio-inspired systems that reproduce functionality and rapid processing of their biological counterpart. In particular, one of the goals of this Special Issue was to explore the possible synergies and interactions of different perspectives for low-level computation up to the system-level brain-like processing. This special issue includes 13 papers. They represent about onehalf of the submitted manuscripts which underwent a peer reviewed process. They can be roughly subdivided in three broad areas, namely: neuromorphic implementation of Spiking Neural Networks; application of bio-inspired hardware to different problems that involve learning; and architectures in emerging technologies such as memristors. We now give a brief summary for each of the papers in this special issue. ‘‘Design of Silicon Brains in the nano-CMOS Era: Spiking Neurons, Learning Synapses and Neural Architecture Optimization’’, by A.S. Cassidy, J. Georgiou and A.G. Andreou, provides an overview setting for this Special Issue. The paper begins with a historical perspective on neuromorphic engineering (Mead, 1990), followed by a fresh view of Marr’s three levels of description, in