SciFiles: Building A Quantum Neural Network

Aug 7, 2017

What is “intelligence”? What is “consciousness”? Is the “mind” separate from the “brain”? Where is the "self" or the "soul"? Some of our deepest questions involve an understanding of the brain.

We can attack these questions by conventional methods: to try to figure out how our own brains work; or we can approach from the other side: to try to build something that can mimic our brains - not just to compute, but to think. This is the goal of my research: I am trying to build a quantum neural network.

Why “quantum”? It is not known whether or not our brains actually are quantum computers, but a computer that uses the quantum mechanical nature of reality would be both much faster and much more powerful than any classical computer could be. For example, it is possible for a system to be in two mutually exclusive states simultaneously. Not: “either on or off, and we don’t know which,” but both on and off, at the same time. This is one of the reasons quantum computing is of such interest: that each different state can be doing a computation, so that two - or eventually, millions - of computations are done at once.

Why “neural network”?

Well, it turns out to be very difficult to design algorithms that exploit the potential power of quantum computing. If, instead, we build a neural network, it can learn, in much the same way that we ourselves learn, not by following a series of logical instructions, but by experience. Instead of a very complicated central processing unit, as in your laptop, we have a very large number of very simple processors—neurons-- that are multiply interconnected. This bypasses the problem of algorithm construction, because the network learns how to do it.

Perhaps, in time, even creativity and consciousness.

Elizabeth is a professor in the physics department of Wichita State University.