Neuromorphic computing: current developments and future challenges

What is it about?

Current computers are based on the “von Neumann architecture,” in which data transfer occurs constantly and operations are performed based on this. Therefore, the speed of the computer is limited by how fast these data can be transferred. Computers in the future, however, need to be faster and more efficient at performing complex tasks. Thus, the rate of data transfer in these computers needs to increase, which can be energy intensive. How can this feature be improved? Turns out, we can learn from our own bodies! The human brain can store and process information very efficiently. One way to design power efficient computers is to model their functioning after the brain (called “neuromorphic computing”). This paper provides an overview of the current progress in this field and challenges that lie ahead. It begins with a discussion of the devices that can be used to mimic the functions of the neurons and synapses of the brain. Neuromorphic circuits and algorithms for replicating the neural networks of the brain are also discussed. Finally, the applications and ethical concerns of this new type of computing are touched upon.

Featured Image

Photo by Milad Fakurian on Unsplash

Photo by Milad Fakurian on Unsplash

Why is it important?

The human brain only uses 20 Watts of power to function. A supercomputer, on the other hand, requires a million times more (i.e., almost 30 Megawatts). The brain's small energy footprint is due to two reasons. First, neurons in the brain exchange information as spikes and energy is only used when a spike occurs. Second, all the information in the brain is stored and processed in the same location. Replicating this functionality would result in faster and more energy efficient computers. KEY TAKEAWAY: Written by experts in the field, this paper covers several topics on neuromorphic computing. It is an exhaustive resource for those interested in the subject.