What are some of the interesting problems whose solutions have been proposed using quantum neural networks?

5

0

I know there are some "quantum versions" of hand-writing recognition algorithms which have been proposed using quantum neural networks. Example: "Recognition of handwritten numerals by Quantum Neural Network with fuzzy features" (J Zhou, 1999). Also, recently by Rebentrost et al.: "A Quantum Hopfield Neural Network" presents an application of their method as a genetic sequence recognizer.

What are some other proposed applications of quantum neural networks whose given solutions provide considerable improvement over the corresponding classical version of the neural network in terms of accuracy? Also, have any of those proposed solutions been programmed/simulated?

Please note that I'm looking for research papers which specifically demonstrate some applications of quantum neural networks and which provide a significant improvement over the corresponding classical neural networks.

Sanchayan Dutta

Posted 2018-05-29T14:59:11.487

Reputation: 14 463

What was the downvote for? – Sanchayan Dutta – 2018-05-29T15:53:23.157

Honestly it was not me. I saw the downote and became curious why it was downvoted, and that's why I came here and then answered. But maybe the reason is this: https://meta.stackexchange.com/questions/158809/why-are-shopping-list-questions-bad

– user1271772 – 2018-05-29T17:22:16.790

1@user1271772 I think it's high time we make a meta post regarding such questions. On a scientific research-level site like this, it is very important that resource-requests like these are allowed. Yes, such list questions are not very suitable on the more layman-sites like Physics SE, but they tend to fit in very well into this site, where researchers often need to request resources corresponding to their research. – Sanchayan Dutta – 2018-05-29T17:39:39.793

@Blue There is a difference between a list question and a resource-request, although it is a bit subtle here. Yes, you may want multiple resources, but why should you explicitly ask for more than one? If someone has a good resource, they can provide it even if there already is one offered. If they have multiple, they are still free to provide them. So, why should a resource request be a list? Also, I'm not exactly sure what being a 'research-level site' has to do with this, nor why Physics isn't such a site and this one apparently is, but that can probably better be discussed elsewhere. – Discrete lizard – 2018-05-29T18:39:52.763

@Discretelizard A list can have one element or infinite elements. So I'm not sure what you're getting at. The list-type questions on Physics SE are quite different from those that get asked here. You'd find answers to most of such questions in standard school or college level textbooks. But surely list questions asking for papers fall in a very different category. – Sanchayan Dutta – 2018-05-30T01:07:35.477

@Blue My point is that as far as I can see, resource requests that happen to be list questions can have the list part removed without harming the question. Since people tend to not like list questions, I see no reason to ask a resource request as a list question. That means that I also see no reason to 'allow' list questions for resource requests. – Discrete lizard – 2018-05-30T07:42:44.157

@Discretelizard Please explain what you mean by "list part removed". I don't think I have even mentioned the word "list" anywhere in the question. – Sanchayan Dutta – 2018-05-30T07:50:46.523

@Blue What I mean is that I see no need to ask for multiple resources. As I said, this was subtle in this question, the usage of the plural form could lead to the interpretation that you're explicitly ask for multiple resources. I didn't interpret the question as such, but it seems user1271772 did. Besides, this was in response to your claims on list questions, which didn't seem to be about this question in particular. – Discrete lizard – 2018-05-30T07:57:22.897

@Discretelizard OK, so according to you: "What are the current alternatives to surface codes, which also take into account next nearest neighbour interactions?" is a list question and shouldn't be allowed? If so, I'd rather disagree. – Sanchayan Dutta – 2018-05-30T08:02:02.550

Let us continue this discussion in chat.

– Discrete lizard – 2018-05-30T08:02:25.277

Answers

1

What are some other proposed applications of quantum neural networks?

Absolutely any application of classical neural networks can be an application of quantum neural networks. There's a lot of examples beyond the two you listed.

Also, have any of those proposed solutions been programmed/simulated?

Yes, for example Ed Farhi of MIT and Hartmut Neven of Google teamed up on a paper where an application was distinguishing digits using QNNs.

user1271772

Posted 2018-05-29T14:59:11.487

Reputation: 8 162

"Absolutely any application of classical neural networks can be an application of quantum neural networks. There's a lot of examples beyond the two you listed." Yes, sure. But I am looking for papers which demonstrate some application. I must have mentioned I was looking for something specifically other than hand-writing recognition and digit-recognition, which are way too common. Anyhow, thanks for your effort. – Sanchayan Dutta – 2018-05-29T17:29:03.087

Also, I'm looking for something that could provide a significant improvement over the classical neural networks. Not all quantum algorithms do. – Sanchayan Dutta – 2018-05-29T17:32:32.507

The Farhi Neven paper prunes the MNIST data so that they only have to distinguish between 2 digits instead of 10. – rrtucci – 2018-05-30T02:51:14.043