The next great leap for computing may be a bit closer with the help of joint efforts between the U.S. government, the private sector — and hundreds of millions of dollars.
And along the way, we might see a benefit for the financial services sector in the form of reduced false positives in fraud detection.
The U.S. Department of Energy said this week that it will spend $625 million over the next five years to develop a dozen research centers devoted to artificial intelligence (AI) and quantum computing. Another $340 million will come from the private sector and academia, bringing Uncle Sam together with the likes of IBM, Amazon and Google to apply the highest of high tech to a variety of verticals and applications.
In an interview with Karen Webster, Dr. Stefan Wörner, global leader for quantum finance and optimization at IBM, said we’re getting closer to crossing the quantum-computing Rubicon from concept to real-world applications.
The basic premise behind quantum computing is that it can tackle tasks with blinding speed and pinpoint accuracy that aren’t possible with “regular” computers. In one famous example, a Google quantum computer late last year reportedly solved an equation in minutes that would have taken the current fastest computers 10,000 years to solve.
IBM is but one marquee tech name developing quantum-computing software, but IBM Quantum last week said that it had made strides in improving hardware and software to boost quantum computing performance.
Similarly, Amazon Web Services recently announced the general availability of Amazon Bracket, which is designed to test and simulate algorithms on simulated quantum computers.
Reducing FIs’ False Positives
Drilling down into the ways in which quantum computing can turbocharge the financial industry, Wörner told Webster that there are three areas of focus: optimization, machine learning (ML) and so-called “Monte Carlo” simulations.
“In each of these areas, there are some algorithms that we think may have the potential to help — not only in terms of speeding up things, but in improving the quality of results,” he said.
That can extend to ML, where classification and segmentation of financial services customers (or “risk ratings”) can be improved, as can fraud detection.
Wörner noted that “one big problem in fraud detection, for example, is credit card fraud. If there is a false positive, then the company has to follow up and find out what is going on, which then leads to manual work” on the part of the financial institution (FI).
And the other side of the coin is reputation loss, as FIs suffer from customer ire when their cards are blocked due to false positives.
The Limits Of Classical Computing
In a “classical” scenario, a typical analysis seeks to take data and form structures within that data, mapping it to a higher “dimensional-feature” space. That means taking one data set and mapping it to another data set that has even more features.
“This is what AI does,” Wörner said. But he added that issues can arise when mapping data to ever-larger feature maps. Regular computing power runs up against limits when it can no longer evaluate feature mapping efficiently.
That’s where quantum computing can step in. As he explained to Webster, quantum computing can take a classical data point of a certain dimension and apply a quantum feature map that maps that point to a very high-dimensional quantum state.
It’s All About Qubits
The rule of thumb for quantum computing all boils down to “qubits” (short for “quantum bits”) that actually power these systems. We’ll need lots of qubits to realize quantum’s potential.
“That way, we can detect more complex structures in the data or improve the quality of the classification,” Wörner said. “And depending on the application [withing financial services], this can help reduce the false positives.”
But he said it might be a few years until we get to those benefits. After all, qubits are at least today subject to “noise” and errors that make them very fragile. Wörner said we can’t perfectly control quantum computers yet. Even though we have 50 qubits available, this noise means that calculations aren’t as precise as they might be otherwise.
Chalk it up to a hardware issue. Wörner said improving the hardware can improve the noise situation and boost the number of qubits across circuits.
However, the investment needed to make this great leap into quantum computing might give banks pause — in part because the return on investment could seem so hard to calculate. But as Wörner told Webster: “This is what makes joining an ecosystem so valuable.”
He pointed to the IBM Q Network, which brings together various stakeholders. Wörner added that the joint effort announced this week between the public and private sectors can serve as tailwinds to expanding computational power. He said IBM is collaborating with banks to develop algorithms and applications via open-source software.
By and large, the advantages to banks as quantum computing become better defined and practical can accrue to the bottom line. The more accurately banks can quantify their risk, the less capital they have to hold as a buffer against that risk (and, say, loan losses). That means FIs can effectively deploy capital elsewhere.
The Importance Of Monte Carlo Simulations
Running Monte Carlo simulations on classical computers would take hours or even days. However, they can be done in seconds with quantum computing and huge data sets.
Better risk analysis also has the advantage of satisfying regulators’ concerns, as FIs can pinpoint risks and address them proactively.
Looking ahead, Wörner said that his firm and the IBM Q Network are looking to make it easier for stakeholders to access quantum computers via the cloud.
The Q Network will ideally look to enable access to quantum resources over the cloud for specific applications without requiring one to be a quantum-computing expert. As Wörner said: “We’re entering the decade of the quantum advantage.”