A review on neuromorphic computing circuits

Abstract

Computational power is measured in terms of the time required for resolutions. The faster, the better. Due to its physical implementation, the conventional computing device used currently in computers is reaching its limit in computational power, and an essential innovation is required in order to break this limit. Neuromorphic circuits, which is inspired by neurology and simulates human biological brain and have higher functional efficiency to consume less energy and perform highly complicated tasks, is introduced and developed. This paper explains the principles of neuromorphic computing, which is representing ions in the biological neural system with electron in the circuit and adopting capacitors and resistors as counterparts of cellular membranes of the neural cells and the ion channel respectively. This paper gives some examples of neuromorphic circuits developed by several different corporations and laboratories. Algorithms mentioned in a certain research is exemplified and explained. Comparison of the difference between conventional and neuromorphic circuits is given to emphasize the advantage of neuromorphic circuits over the conventional ones. Several possible applications in a range of fields are also provided to depict the future prospect of this technology, including artificial intelligence, statistical calculation and information analysis. The conclusion is that the neuromorphic computers will replace the conventional Von Neumann computers, boosting the further development in computing power, breaking its limit.