[ad_1]
IBM has discovered how to ignore noisy qubits and execute machine learning algorithms in quantum function spaces. Eureka cadabra! The age of quantum algorithms is upon us.
A team of IBM researchers, alongside MIT scientists and Oxford, created a pair of quantum classification algorithms that they then experimentally implemented on a hybrid system using a 2-qubit quantum computer and a classical superconductor. Roughly, they have shown that quantum computers can offer advantages in machine learning, unlike conventional computers.
According to the researchers' paper:
Here, we experimentally propose and implement two quantum algorithms on a superconducting processor. A key element in both methods is the use of the quantum state space as a function space. The use of a quantum-enhanced feature space that is only efficiently available on a quantum computer provides a possible path to a quantum advantage. The algorithms solve a problem of supervised learning: the construction of a classifier.
Another way of saying it: we now have a road map for a quantum advantage in machine learning. This is the point where the capacity of a quantum system to execute / optimize algorithms exceeds that of a conventional computer. We are not there yet, as the IBM research blog points out:
Our research does not yet demonstrate the Quantum advantage, as we have minimized the magnitude of the problem based on our current hardware capabilities, using only two qubits of quantum computing capability, which can be simulated on a classic computer … What we showed is promising. the path to follow.
TNW Kristan Temme, physicist from IBM Research and co-author of the team's white paper, and Bob Sutor, vice president of IBM Q Strategy, spoke about Reese's mix of knowledge and machine learning: quantum. Temme told us that the team designed the experiment to work with noisy systems today: "These are essentially algorithms that should apply to a device that has no fault tolerance," he explains.
This is important because, in the current state of affairs, one of the main obstacles to the usefulness of quantum computers outside of laboratories is the problem of decoherence, which is essentially a manifestation of quantum noise (more details here). ).
The idea here is not to wait until quantum hardware is perfect in a decade or two before you begin to understand how to develop and program these systems. IBM's work shows how conventional and quantum computers will work together to solve problems.
And speaking of working together: IBM has opened the algorithms. If you're wondering why a big tech company would do this on Earth, you're not alone. We asked Sutor, who told us:
We are doing everything we can to help people take control … We have learned a lot about open source over the years. Open source is an essential way to develop software.
Temme added: "We hope a lot of people will engage in algorithms."
To that end, you can try an interesting demonstration of the algorithms here. No quantum physics or computer skills are required. For those who want to go a step further: IBM has provided them to Qiskit Aqua, an open source library of quantum algorithms for developers and researchers to use with quantum computers accessed from IBM's cloud.
We are still in the early days of quantum computing, as recently stated by Bob Wisnieff, IBM's CTO in Quantum Computing, at TNW:
Imagine that everyone in the 60s has five to ten years to explore the hardware and programming of the mainframe while it was still essentially a prototype. This is where we are with quantum computing.
IBM's latest research paves the way for both quantum computers and machine learning. We can not wait to see what is on the other side of the benefit of quantum learning.
Want to know more about AI than the titles? Discover our Track machine learners at TNW2019.
Plunge into the heart of the craziest plot theories of Bitcoin
[ad_2]
Source link