Machine Learning Shows Potential For Improving Quantum Computers

[Lucian Armasu]

Scientists from the University of Bristol used machine learning techniques applied to a silicon-based quantum photonics processor to advanced their knowledge of quantum computing systems.

Machine Learning Shows Potential For Improving Quantum Computers : Read more

 

[Ford__]

Good overview article and this topic is fascinating. With the eventual development of quantum computers (and I do believe it is more of a question of "when" than "if") we will be able to do some really wacky things, as suggested in works like "On Computer Simulated Universes". It seems that we may finally be on the cusp of another technological revolution...

 

[vern72]

Yup, can't wait for the day when Skynet becomes self aware.

 

[bit_user]

we will be able to do some really wacky things, as suggested in works like "On Computer Simulated Universes".

To simulate the universe, you'd need a computer at least the size of one.

It seems that we may finally be on the cusp of another technological revolution...

Yes, it is a revolution. But you'd do well to understand the kinds of problems they're adept at solving.

They're good at problems that are difficult for classical computers, but not at the types of things that classical computers do well. They also take a bit of time to find an optimal solution. All very different than your PC.

 

[bit_user]

Yup, can't wait for the day when Skynet becomes self aware.

The article didn't explain this well, but the way these quantum computers would be applied to machine learning is to train a neural net to do some task. Like recognizing characters or translating speech. But they're all sort of pattern recognition tasks.

The thing is, the quantum computer wouldn't be running the neural net. It wouldn't be thinking. It would train a neural net, using a large pool of input data. Then the neural net would then remain fairly static, as it's applied to more data for whatever task it was trained to do.