Tag Archives: mechanics
#438982 Quantum Computing and Reinforcement ...
Deep reinforcement learning is having a superstar moment.
Powering smarter robots. Simulating human neural networks. Trouncing physicians at medical diagnoses and crushing humanity’s best gamers at Go and Atari. While far from achieving the flexible, quick thinking that comes naturally to humans, this powerful machine learning idea seems unstoppable as a harbinger of better thinking machines.
Except there’s a massive roadblock: they take forever to run. Because the concept behind these algorithms is based on trial and error, a reinforcement learning AI “agent” only learns after being rewarded for its correct decisions. For complex problems, the time it takes an AI agent to try and fail to learn a solution can quickly become untenable.
But what if you could try multiple solutions at once?
This week, an international collaboration led by Dr. Philip Walther at the University of Vienna took the “classic” concept of reinforcement learning and gave it a quantum spin. They designed a hybrid AI that relies on both quantum and run-of-the-mill classic computing, and showed that—thanks to quantum quirkiness—it could simultaneously screen a handful of different ways to solve a problem.
The result is a reinforcement learning AI that learned over 60 percent faster than its non-quantum-enabled peers. This is one of the first tests that shows adding quantum computing can speed up the actual learning process of an AI agent, the authors explained.
Although only challenged with a “toy problem” in the study, the hybrid AI, once scaled, could impact real-world problems such as building an efficient quantum internet. The setup “could readily be integrated within future large-scale quantum communication networks,” the authors wrote.
The Bottleneck
Learning from trial and error comes intuitively to our brains.
Say you’re trying to navigate a new convoluted campground without a map. The goal is to get from the communal bathroom back to your campsite. Dead ends and confusing loops abound. We tackle the problem by deciding to turn either left or right at every branch in the road. One will get us closer to the goal; the other leads to a half hour of walking in circles. Eventually, our brain chemistry rewards correct decisions, so we gradually learn the correct route. (If you’re wondering…yeah, true story.)
Reinforcement learning AI agents operate in a similar trial-and-error way. As a problem becomes more complex, the number—and time—of each trial also skyrockets.
“Even in a moderately realistic environment, it may simply take too long to rationally respond to a given situation,” explained study author Dr. Hans Briegel at the Universität Innsbruck in Austria, who previously led efforts to speed up AI decision-making using quantum mechanics. If there’s pressure that allows “only a certain time for a response, an agent may then be unable to cope with the situation and to learn at all,” he wrote.
Many attempts have tried speeding up reinforcement learning. Giving the AI agent a short-term “memory.” Tapping into neuromorphic computing, which better resembles the brain. In 2014, Briegel and colleagues showed that a “quantum brain” of sorts can help propel an AI agent’s decision-making process after learning. But speeding up the learning process itself has eluded our best attempts.
The Hybrid AI
The new study went straight for that previously untenable jugular.
The team’s key insight was to tap into the best of both worlds—quantum and classical computing. Rather than building an entire reinforcement learning system using quantum mechanics, they turned to a hybrid approach that could prove to be more practical. Here, the AI agent uses quantum weirdness as it’s trying out new approaches—the “trial” in trial and error. The system then passes the baton to a classical computer to give the AI its reward—or not—based on its performance.
At the heart of the quantum “trial” process is a quirk called superposition. Stay with me. Our computers are powered by electrons, which can represent only two states—0 or 1. Quantum mechanics is far weirder, in that photons (particles of light) can simultaneously be both 0 and 1, with a slightly different probability of “leaning towards” one or the other.
This noncommittal oddity is part of what makes quantum computing so powerful. Take our reinforcement learning example of navigating a new campsite. In our classic world, we—and our AI—need to decide between turning left or right at an intersection. In a quantum setup, however, the AI can (in a sense) turn left and right at the same time. So when searching for the correct path back to home base, the quantum system has a leg up in that it can simultaneously explore multiple routes, making it far faster than conventional, consecutive trail and error.
“As a consequence, an agent that can explore its environment in superposition will learn significantly faster than its classical counterpart,” said Briegel.
It’s not all theory. To test out their idea, the team turned to a programmable chip called a nanophotonic processor. Think of it as a CPU-like computer chip, but it processes particles of light—photons—rather than electricity. These light-powered chips have been a long time in the making. Back in 2017, for example, a team from MIT built a fully optical neural network into an optical chip to bolster deep learning.
The chips aren’t all that exotic. Nanophotonic processors act kind of like our eyeglasses, which can carry out complex calculations that transform light that passes through them. In the glasses case, they let people see better. For a light-based computer chip, it allows computation. Rather than using electrical cables, the chips use “wave guides” to shuttle photons and perform calculations based on their interactions.
The “error” or “reward” part of the new hardware comes from a classical computer. The nanophotonic processor is coupled to a traditional computer, where the latter provides the quantum circuit with feedback—that is, whether to reward a solution or not. This setup, the team explains, allows them to more objectively judge any speed-ups in learning in real time.
In this way, a hybrid reinforcement learning agent alternates between quantum and classical computing, trying out ideas in wibbly-wobbly “multiverse” land while obtaining feedback in grounded, classic physics “normality.”
A Quantum Boost
In simulations using 10,000 AI agents and actual experimental data from 165 trials, the hybrid approach, when challenged with a more complex problem, showed a clear leg up.
The key word is “complex.” The team found that if an AI agent has a high chance of figuring out the solution anyway—as for a simple problem—then classical computing works pretty well. The quantum advantage blossoms when the task becomes more complex or difficult, allowing quantum mechanics to fully flex its superposition muscles. For these problems, the hybrid AI was 63 percent faster at learning a solution compared to traditional reinforcement learning, decreasing its learning effort from 270 guesses to 100.
Now that scientists have shown a quantum boost for reinforcement learning speeds, the race for next-generation computing is even more lit. Photonics hardware required for long-range light-based communications is rapidly shrinking, while improving signal quality. The partial-quantum setup could “aid specifically in problems where frequent search is needed, for example, network routing problems” that’s prevalent for a smooth-running internet, the authors wrote. With a quantum boost, reinforcement learning may be able to tackle far more complex problems—those in the real world—than currently possible.
“We are just at the beginning of understanding the possibilities of quantum artificial intelligence,” said lead author Walther.
Image Credit: Oleg Gamulinskiy from Pixabay Continue reading
#436261 AI and the future of work: The prospects ...
AI experts gathered at MIT last week, with the aim of predicting the role artificial intelligence will play in the future of work. Will it be the enemy of the human worker? Will it prove to be a savior? Or will it be just another innovation—like electricity or the internet?
As IEEE Spectrum previously reported, this conference (“AI and the Future of Work Congress”), held at MIT’s Kresge Auditorium, offered sometimes pessimistic outlooks on the job- and industry-destroying path that AI and automation seems to be taking: Self-driving technology will put truck drivers out of work; smart law clerk algorithms will put paralegals out of work; robots will (continue to) put factory and warehouse workers out of work.
Andrew McAfee, co-director of MIT’s Initiative on the Digital Economy, said even just in the past couple years, he’s noticed a shift in the public’s perception of AI. “I remember from previous versions of this conference, it felt like we had to make the case that we’re living in a period of accelerating change and that AI’s going to have a big impact,” he said. “Nobody had to make that case today.”
Elisabeth Reynolds, executive director of MIT’s Task Force on the Work of the Future, noted that following the path of least resistance is not a viable way forward. “If we do nothing, we’re in trouble,” she said. “The future will not take care of itself. We have to do something about it.”
Panelists and speakers spoke about championing productive uses of AI in the workplace, which ultimately benefit both employees and customers.
As one example, Zeynep Ton, professor at MIT Sloan School of Management, highlighted retailer Sam’s Club’s recent rollout of a program called Sam’s Garage. Previously customers shopping for tires for their car spent somewhere between 30 and 45 minutes with a Sam’s Club associate paging through manuals and looking up specs on websites.
But with an AI algorithm, they were able to cut that spec hunting time down to 2.2 minutes. “Now instead of wasting their time trying to figure out the different tires, they can field the different options and talk about which one would work best [for the customer],” she said. “This is a great example of solving a real problem, including [enhancing] the experience of the associate as well as the customer.”
“We think of it as an AI-first world that’s coming,” said Scott Prevost, VP of engineering at Adobe. Prevost said AI agents in Adobe’s software will behave something like a creative assistant or intern who will take care of more mundane tasks for you.
“We need a mindset change. That it is not just about minimizing costs or maximizing tax benefits, but really worrying about what kind of society we’re creating and what kind of environment we’re creating if we keep on just automating and [eliminating] good jobs.”
—Daron Acemoglu, MIT Institute Professor of Economics
Prevost cited an internal survey of Adobe customers that found 74 percent of respondents’ time was spent doing repetitive work—the kind that might be automated by an AI script or smart agent.
“It used to be you’d have the resources to work on three ideas [for a creative pitch or presentation],” Prevost said. “But if the AI can do a lot of the production work, then you can have 10 or 100. Which means you can actually explore some of the further out ideas. It’s also lowering the bar for everyday people to create really compelling output.”
In addition to changing the nature of work, noted a number of speakers at the event, AI is also directly transforming the workforce.
Jacob Hsu, CEO of the recruitment company Catalyte spoke about using AI as a job placement tool. The company seeks to fill myriad positions including auto mechanics, baristas, and office workers—with its sights on candidates including young people and mid-career job changers. To find them, it advertises on Craigslist, social media, and traditional media.
The prospects who sign up with Catalyte take a battery of tests. The company’s AI algorithms then match each prospect’s skills with the field best suited for their talents.
“We want to be like the Harry Potter Sorting Hat,” Hsu said.
Guillermo Miranda, IBM’s global head of corporate social responsibility, said IBM has increasingly been hiring based not on credentials but on skills. For instance, he said, as much as 50 per cent of the company’s new hires in some divisions do not have a traditional four-year college degree. “As a company, we need to be much more clear about hiring by skills,” he said. “It takes discipline. It takes conviction. It takes a little bit of enforcing with H.R. by the business leaders. But if you hire by skills, it works.”
Ardine Williams, Amazon’s VP of workforce development, said the e-commerce giant has been experimenting with developing skills of the employees at its warehouses (a.k.a. fulfillment centers) with an eye toward putting them in a position to get higher-paying work with other companies.
She described an agreement Amazon had made in its Dallas fulfillment center with aircraft maker Sikorsky, which had been experiencing a shortage of skilled workers for its nearby factory. So Amazon offered to its employees a free certification training to seek higher-paying work at Sikorsky.
“I do that because now I have an attraction mechanism—like a G.I. Bill,” Williams said. The program is also only available for employees who have worked at least a year with Amazon. So their program offers medium-term job retention, while ultimately moving workers up the wage ladder.
Radha Basu, CEO of AI data company iMerit, said her firm aggressively hires from the pool of women and under-resourced minority communities in the U.S. and India. The company specializes in turning unstructured data (e.g. video or audio feeds) into tagged and annotated data for machine learning, natural language processing, or computer vision applications.
“There is a motivation with these young people to learn these things,” she said. “It comes with no baggage.”
Alastair Fitzpayne, executive director of The Aspen Institute’s Future of Work Initiative, said the future of work ultimately means, in bottom-line terms, the future of human capital. “We have an R&D tax credit,” he said. “We’ve had it for decades. It provides credit for companies that make new investment in research and development. But we have nothing on the human capital side that’s analogous.”
So a company that’s making a big investment in worker training does it on their own dime, without any of the tax benefits that they might accrue if they, say, spent it on new equipment or new technology. Fitzpayne said a simple tweak to the R&D tax credit could make a big difference by incentivizing new investment programs in worker training. Which still means Amazon’s pre-existing worker training programs—for a company that already famously pays no taxes—would not count.
“We need a different way of developing new technologies,” said Daron Acemoglu, MIT Institute Professor of Economics. He pointed to the clean energy sector as an example. First a consensus around the problem needs to emerge. Then a broadly agreed-upon set of goals and measurements needs to be developed (e.g., that AI and automation would, for instance, create at least X new jobs for every Y jobs that it eliminates).
Then it just needs to be implemented.
“We need to build a consensus that, along the path we’re following at the moment, there are going to be increasing problems for labor,” Acemoglu said. “We need a mindset change. That it is not just about minimizing costs or maximizing tax benefits, but really worrying about what kind of society we’re creating and what kind of environment we’re creating if we keep on just automating and [eliminating] good jobs.” Continue reading