Quantum Computers Can Now Do Machine Learning – We Just Need To Build One That Works

IBM has come up with a way to use quantum computers to improve machine learning algorithms, even though we don’t have anything approaching a quantum computer yet.

The tech giant developed and tested a quantum algorithm for machine learning with scientists from Oxford University and MIT, showing how quantum computers will be able to map data at a far more sophisticated level than any classical computer.

Somewhat ironically, the testing was done using modelling of only two qubits simulated on a classical computer, because that’s the current hardware capability available. There are no quantum computers because qubits can’t stay in an entangled state for more than a few hundred microseconds, even in carefully controlled laboratory conditions. They break down into decoherence and can no longer be used to perform calculations in parallel, the feature of quantum computing that will give it awesome processing power.

(Credit: Getty)GETTY

This is the common problem for all quantum computing hopefuls, but IBM has decided not to wait until the problem is solved to start programming the applications that will use it.

“We are still far off from achieving Quantum Advantage for machine learning – the point at which quantum computers surpass classical computers in their ability to perform AI algorithms,” said IBM in a research blogpost.

“Yet the feature mapping methods we’re advancing could soon be able to classify far more complex datasets than anything a classical computer could handle. What we’ve shown is a promising path forward.”

Machine learning algorithms get smarter the more (accurate) data you feed them. So higher levels of processing power to chew through much huger datasets will bring us ever more intelligent computer programs. This algorithm shows how quantum computing can improve feature mapping, a trick computer programs use to analyse data in a completely different way to human beings.

Basically, this is where the program breaks down the data to get to information at a much more granular level. For example, both classical and quantum computers could break down a picture by pixels and then place them into a grid based on each pixel’s colour value. Quantum computers will essentially be able to break down data even more finely and build highly complex feature maps in high-dimensional space, getting at aspects of data that classical computers can’t even dream of.

“The goal is to use quantum computers to create new classifiers that generate more sophisticated data maps. In doing that, researchers will be able to develop more effective AI that can, for example, identify patterns in data that are invisible to classical computers,” IBM explained.

“We’ve developed a blueprint with new quantum data classification algorithms and feature maps. That’s important for AI because, the larger and more diverse a data set is, the more difficult it is to separate that data out into meaningful classes for training a machine learning algorithm. Bad classification results from the machine learning process could introduce undesirable results; for example, impairing a medical device’s ability to identify cancer cells based on mammography data.”

Many of the algorithms that IBM is developing will be available as part of its open-source library Qiskit Aqua for researchers and other developers.

[“source=forbes”]