Training a deep neural net on a quantum computer
3 comments
1. There are hundreds of hypotheses about our brains. Neuroscientists barely understand our brain, so conjecture about what our brains do is just that - conjecture.
2. Quantum computers basically only offer a runtime speedup for optimization. It'll probably help train neural nets faster, but that's not going to help you in any other area.
2. Quantum computers basically only offer a runtime speedup for optimization. It'll probably help train neural nets faster, but that's not going to help you in any other area.
Quantum computers and AI seem like they're made for each other, because both depend on optimization algorithms. Now, whether our current AI algorithms can be sped-up greatly by quantum computers, or we'll need to find new AI algorithms that are sped-up by them, we'll have to see. I'm sure Google will try to port DeepMind to its own universal quantum computer in the next few years.
I agree. Bringing some funky extreme non-linearity into A.I training would be interesting. However, the models would probably be very different than what we have today.
Who else are truly building quantum computers other than D-Wave and IBM?
Who else are truly building quantum computers other than D-Wave and IBM?
Roger Penrose has hypothesized that consciousness could be a result of quantum effects in neurons..
http://rationalwiki.org/wiki/Quantum_consciousness