Following D-Wave’s announcement of Leap 2, a brand new model of its quantum cloud service for constructing and deploying quantum computing purposes, VentureBeat had the chance to sit down down with Murray Thom, D-Wave’s VP of software program and cloud companies. We naturally talked about Leap 2, together with the enhancements the corporate hopes it’s going to carry for companies and builders. But we additionally mentioned the enterprise purposes D-Wave has already seen so far.
Quantum computing leverages qubits to carry out computations that will be rather more troublesome, or just not possible, for a classical pc. Based in Burnaby, Canada, D-Wave was the primary firm to promote industrial quantum computer systems, that are constructed to make use of quantum annealing. Applications embrace every part from cryptography and optimization to machine studying and supplies science. In reality, D-Wave has a webpage devoted to quantum computing applications together with airline scheduling, election modeling, quantum chemistry simulation, automotive design, preventative well being care, logistics, and extra.
Thom defined that D-Wave has seen success significantly with optimization and machine studying use instances. And he has the information to again it up: D-Wave’s buyer purposes are about 50% optimization, 20% AI and ML, 10% supplies science, and 20% different.
Optimization vs. machine studying
Optimization purposes main the pack is sensible as a result of they’re at the moment largely solved utilizing brute drive and uncooked computing energy. If quantum computer systems can rapidly see all of the doable options, an optimum resolution can turn into obvious extra rapidly. “Optimization stands out because it’s much more intuitive and easier to grasp,” Thom added.
Another cause optimization is forward comes all the way down to the events.
“The community of people who can incorporate optimization and robust optimization is much, much larger,” Thom defined. “The machine learning community — the congruence between the technology and the needs are very technical; they’re only applicable to statisticians. And there’s a much smaller community of statisticians in the world than there are of programmers.”
In specific, the complexity of incorporating quantum computing into the machine studying workflow presents an impediment. “For machine learning practitioners and researchers, it’s very easy to figure out how to program the system. Fitting that into a machine learning workflow is more challenging because machine learning programs are becoming quite multifaceted,” he mentioned. “But our teams in the past have published a lot of research on how to incorporate it in a training workflow that makes sense.”
Indeed, Thom famous that ML practitioners at the moment need another person to deal with the quantum computing half: “When I’ve gone out and talked to the machine learners, they’re looking for somebody else to do the legwork of building the frameworks up to the extensions and showing that it can fit.”
Adoption will take time
Nonetheless, Thom believes quantum computing and machine studying are “extremely well matched. The features the technology has and the needs of the field are very close.”
“It’s something I think is going to be a very productive use of the technology in the future because there’s so many aspects of what the quantum computers can do in terms of the probabilistic sampling,” Thom continued. “For optimization, the probabilistic sampling is like ‘oh, I can do robust optimization with that.’ But for machine learning it’s essential for what you need to do. It’s very hard to reproduce that with a classical computer and you get it natively from the quantum computer. So those features can’t be accidental. It’s just that it’s going to take time for the community to find the right methods for incorporating it and then for the technology to insert into that space productively.”
We’ve seen AI and quantum computing collide earlier than, however this can be a a lot larger nod of approval. Still, D-Wave is only one quantum computing firm. We reached out to a couple others to seek out out if that they had seen related numbers.
Rigetti doesn’t have a breakdown of use instances its prospects fall into. IonQ defined that whereas it’s serving to corporations most notably throughout power, pharma, and manufacturing, it additionally doesn’t have an actual breakdown of shoppers by use case. IBM didn’t reply in time for publishing. D-Wave says that quantum computing and machine studying are a very good match. If you suppose in any other case, let us know.