Check out this news item:
Quantum Machine Learning for Big-Data, by R. Colin Johnson (eetimes.com, 7/26/2013)
Excerpts in boldface:
“What can we use quantum computers for?” Lloyd said in his ICQT session. “I would like to propose the first quantum app, or q-app, which I call ‘Quantum Machine Learning for Big Quantum Data.’ ”
Hey, how about mentioning your 2 collaborators. I believe their names are Minion A and Minion B.
The first quantum app? I don’t think so. What about Shor and Grover? The Lloyd-app is not even the first one that does machine learning. Long before the Lloyd vaporware, D-Wave/Google and other researchers (including myself) have been writing papers and software for doing AI with a QC.
Lloyd has tried to get commercial funding to develop his q-app, but has so far failed to convince any venture capitalist to fund his project. The reason, he claims, is that his q-app insures that the search engine would not be able to store the user’s queries to add to the reams of information they already store about each of their users.
Maybe the reason VCs don’t bite is because he has no business experience or real interest in running a business, except to get a vanity plate for his car. Maybe they don’t trust him because he exaggerates too much. A case of crying wolf one too many times?
And after losing control of the technology that he claims to have invented, which he says D-Wave is currently using without paying him royalties, this time he has patented his Quantum Machine Learning q-app.
I don’t know what he is talking about. The consensus seems to be that Farhi, Goldstone, Gutmann, and Sipser invented adiabatic quantum computers. Maybe he thinks he invented Josephson junctions or squids?
Lloyd has tested his algorithm — in theory and on a small scale — using a conventional supervised machine learning algorithm operating on his q-bit space that shrinks classical data sets exponentially. And he claims the process worked so well that even the largest data set of all — every bit in the universe, which he describes in his book Programming the Universe, would only require a quantum computer with 300 q-bits to query in real-time.
How about quantum decoherence/error correction and the minor problem of storing the universe inside a subset of itself? I think he should read this brief quote by Feynman.
If you want to learn more about Seth Lloyd, you might want to skim through his book “Programming the Universe” or at least read the hilarious one star comments about the book at Amazon.com. You might also want to check out one of his numerous YouTube videos. Personally, I find those videos Très Void.
And don’t forget to check out the 2 arxiv papers behind this story:
Quantum support vector machine for big feature and big data classification, by
Patrick Rebentrost, Masoud Mohseni, Seth Lloyd (http://arxiv.org/abs/1307.0471)
Quantum algorithms for supervised and unsupervised machine learning, by
Seth Lloyd, Masoud Mohseni, Patrick Rebentrost (http://arxiv.org/abs/1307.0411)
Don’t get me wrong, I think using QCs to do machine learning and AI is a great idea, but I don’t think the work of Lloyd et al is either the first or the final word on this subject.