Archived

This topic is now archived and is closed to further replies.

Prozak

Using Quantum Computers in AI

Recommended Posts

I''ve been reading a bit on Quantum computers, and how they use qubits. Aparently, a qubit can store any number of possible results for a certain problem. Could we construct a neuron out of a qubit? Well, maybe not one, but a few. If we start seing AI at this level, then any being using a quBrain with the cubic area of my thumg would be smarter than the sum of all the IQs of my ancesters for the past 20.000 years... (please disregard the fact that any of this technology is years into the future and feel free to speculate). A cookie for your thoughts?

[Hugo Ferreira][Positronic Dreams][Colibri 3D Engine][Entropy HL2 MOD][Yann L.][Enginuity]
Stop reading my signature and click it!

Share this post


Link to post
Share on other sites
I have only a marginal understanding of quantum computing, but I would imagine that quantum computers would make powerful combinatoric optimizers. Many problems which are difficult on serial computers (search and optimization, especially) might be a lot faster on a quantum computer.

-Predictor


[edited by - Predictor on November 18, 2003 1:03:24 PM]

Share this post


Link to post
Share on other sites
Intelligence is subjective. Even with a computer 100000 times more powerful than the human brain you would still need to program that machine to act in an ''intelligent'' way. You can''t just expect a chess playing computer to wake up one day and start talking to it''s operator about the weather.

That said, if you could somehow simulate physics accurately enough to model an already existing life from at the chemical level (or whatever resolution is requird to mimic life), maybe we could see a real ''AI'' living in a real ''VR''.

AFAIK somebody did run a simulation of an actual biological virus in this way. Anyone have details on this?

Will

Share this post


Link to post
Share on other sites
quote:
Original post by pentium3id
Could we construct a neuron out of a qubit?

No.

A qubit can be any number of results at once, but you can only use one. As they stand now, this would be useless in neurons.

If anything, quantum computers will make training more simple. With a proper algorithm to back it, you could theoretically pass it several training samples at a time. But I imagine would only be in special circumstances which would likely be useless in a real situation. But this all rests on finding an algorithm that does this, and most scientists are focusing on more useful targets.

CM

Share this post


Link to post
Share on other sites
Quantum Computers are good at certain types of tasks (as previously mentioned). Just to expand a bit on what has been said. Quantum computers can execute a operation on a whole bunch of values at once (all possible values in most cases), you can only get one result out of it though. So they don''t really help you speed up typical algorithms. However, they can tell you something about the set of results from running that operation on all possible inputs. For example, if you wish to find the minimum value of the results, quantum computers are quite good at that. Or if you wish to check if all the results are equal, or if they all have the same parity, etc. Qunatum computers can do that in just one operation, whereas a classical computer would have to run that operation multiple times, once for each possible input.

Share this post


Link to post
Share on other sites