News Researchers Demonstrate First ‘Poor Man’s Qubit’ Probabilistic Computing Hardware

bit_user

Polypheme
Ambassador
This is neat. It definitely reminds me of a neural network, in the way they apparently rely on a noisy input and then some sort of feedback mechanism to converge on the solution for a given, known output.

I also wonder how well they'll scale. I mean, to the extent that they just depend on probabilistic coverage of the solution space, it seems like efficiency would be inversely-related to the number of inputs. I guess not terribly dissimilar to how a large neural network takes a lot more time & energy to train than a small one.

My analogies might be off, as I'm too cheap/lazy to buy the original Nature article.