Thursday, May 24

AI&ML quantum talk: Accenture Labs

LinkedIn Google+ Pinterest Tumblr +

Accenture Labs is using and deploying quantum algorithms. Carl Dukatz, senior technology principal and Malek Ben Salem, senior researcher in the cybersecurity R&D Lab talk to MarketBrains about the opportunities and risks for financial services.

MarketBrains: The quantitative trading community is becoming more aware of quantum computers and how it might apply to finance. What does that look like on your end?

Carl Dukatz, Senior Technology Principal, Accenture Labs

Carl Dukatz: We’re tracking and understanding what is going on with the hardware, but we’re also tracking and understanding what’s going on with different software toolkits, and specifically different algorithms that are coming out within those toolkits.

For us, it’s helpful to understand the kits that are available today and how they will grow in the future and to map those kits and algorithms to business use cases or real use cases that exist already.

 

“I don’t think that there’s anyone in the financial services industry who is not aware that quantum is coming along”

 

MB: And what does that look like right now?

CD:  The easiest thing to simulate in the quantum space, the most direct correlation, is in chemistry. People think that is the most likely space where we can map what quantum is doing to something tangible we are doing in the world today using universal model (or gate model) quantum computers.

We can model molecules more effectively using quantum computation instead of classical computation, and that will likely be the first thing to demonstrate quantum supremacy. Quantum supremacy meaning the first use case where we show quantum computers are truly faster or more useful than a classical technique or machine.

MB: How many qubits are we talking about to do this?

CD: Researchers have shown that even a 5 qubit machine can model molecules, so we know that these types of problems can be done. 17 qubits is an interesting number because it allows us to do more robust activities with error correction.

The area where people are really getting excited is if you can get to around the 50+ qubit range. The reason that is an important number is that the best that we can do with all of our other technologies is 56 qubits.

If we can do more than 56 qubits, then we will be able to simulate some of these molecules that are currently not tractable with classical computers. However, classical techniques are coming out frequently that continue to push quantum supremacy back, so I wouldn’t be surprised to see the numbers change.

 

“No one wants to be disrupted and that’s why everyone is pushing very hard to be a part of quantum, they stand to lose the most”

 

MB: What is something practical that can be done with quantum computers today?

CD: A specific example is in pharmaceuticals with Biogen, where we used the quantum annealer to compare small molecules within the pharmaceutical space. Biogen is a research firm specifically looking at neurological diseases, for which you need compounds that both can help address the disease as well as penetrate the blood brain barrier in order to get to the nervous system where they can be effective.

They are looking for new molecules that match multiple types of features, and so this molecular comparison use case is very interesting for them because it gives them a new powerful tool to help them find and create the drugs.

 

“…a lot of the foundational investors in these quantum start-ups that are focusing on algorithms have come out of the financial services industry”

 

MB: Are you working with firms in quantum machine learning for finance?

CD:  For the use cases, what we’re seeing is that with the quantum annealing approach we can apply that to some business problems today.

The obvious one in investment finance is portfolio optimization. If you can take in an additional number of features that no one else can compute at this point in time, then you can effectively come to conclusions about portfolios and make recommendations that no one else can.

Also, a lot of the foundational investors in these quantum start-ups that are focusing on algorithms have come out of the financial services industry.

I don’t think that there’s anyone in the financial services industry who is not aware that quantum is coming along. The industry itself has been typically an early adopter of new forms of compute, especially ones that could lead to better optimization of trade or analysis of the indicators within the financial industry or markets.

The key interest is if quantum provides a mechanism for analyzing data in a way that they currently can’t analyze, or take in more factors than what is currently computationally possible in order to create predictions of the market or trade.

Companies that have that capability can effectively move the bar forward very far in transactions that other people aren’t necessarily capable of doing, so if you can make decisions much faster and more strategically in that space then you stand to create a lot of revenue.

No one wants to be disrupted and that’s why everyone is pushing very hard to be a part of quantum, they stand to lose the most.

 

“…there will be a window of time when a sophisticated attacker would be able to decrypt everything we collect today”

 

MB: There’s another aspect of quantum computing that is making headlines, and that’s around cyber security. One of the problems that quantum computers solve will mean that the world’s current encryption methods will become obsolete. Intelligence agencies are already issuing guidance about this, so can the largest financial firms be far behind in these concerns?

Malek Ben Salem, Senior Researcher, Cybersecurity R&D Lab, Accenture Labs

Malek Ben Salem: The risk is that all of the existing cryptographic methods that are based on complicated factoring algorithms will be broken.

We have some assumptions about how difficult it is to factor real big integer numbers using existing classical computing methods.

What quantum computing will be able to do is factor those numbers easily and therefore will be able to identify the cryptographic keys that are being generated. That breaks the entire cryptosystem and it breaks the entire assumptions about the security and confidentiality of cryptosystems.

The intelligence community is currently working on new quantum resistant crypto algorithms, and there are some versions which we have not fully proven to be inherently secure against attack by a quantum computer. Other algorithms are provably secure, but not scalable.

For companies and for society, the challenge is to deploy those algorithms. If you’re changing an entire cryptosystem deployment, it will take probably a decade.

By the time we will be done deploying them, quantum could have solved already the problem of factoring, efficiently using Shor’s factoring algorithm, therefore there will be a window of time when a sophisticated attacker would be able to decrypt everything we collect today.

It’s not a reality but it’s a real threat.

This interview is edited and condensed

Share.

About Author

Leave A Reply