Jan 16, 2018

Acceleration. It’s the challenge Arthur Redfern focuses on every day as manager of our Machine Learning Lab. He’s working on sophisticated methods to make machines smarter with algorithms, software and optimized processors.

But he has been accelerating innovation at our company from the day he walked in the door. It started with his job interview in late 1999, when he spent an hour discussing his doctoral work in nonlinear systems at Georgia Tech. A senior technology leader in the audience thought that Arthur’s ideas could be applied to a broadband communication problem he was trying to solve. The two sat down to discuss the problem and, a few minutes later, had a solution that resulted in an IEEE journal paper and a U.S. patent.

"It was the greatest interview I’ll ever have in my life," Arthur said.

He joined our company in 2000 and spent most of the following decade working on physical layer communication systems. Following that, he worked on signal processing for analog systems developing solutions for better analog-to-digital converters, touch-screen controllers and audio speakers. In 2013, he shifted gears again and formed what became the Machine Learning Lab.

"Arthur recognized the impact machine learning was going to have before most of us did," said his supervisor, Sanjive Agarwala. "And because the implications are such a game-changer, he’s been working with other organizations throughout the company to tailor machine learning to their areas."

Innovation across multiple disciplines

Arthur is a prolific inventor, with 28 patents to his credit. His approach to research combines a big-picture system view with formal bottoms-up math. "I think it makes me a better technology manager if I have a detailed understanding of everything from the application to the algorithms to the hardware," he said.

Under his guidance, the Machine Learning Lab is developing faster hardware and easier-to-use software libraries that enable customers to design their own sophisticated artificial intelligence (AI) systems.

"In machine learning, we love generality," said Tarek Lahlou, a systems engineer in the lab. "There’s a new ’best algorithm’ every week in AI, so we’ve made something that can handle whatever you throw at it for the next five years."

Serious about speed

Once he gets started on a problem, or explaining an innovation, he doesn’t slow down.

"He’s so passionate, and has so much information, that it can be overwhelming to people on the other side of the table," Sanjive said. "But we know that he will never give up, and he gets people to embrace what he’s trying to do."

You can see that passion at work when Arthur recruits talented engineers from top universities. Sanjive credits him with bringing some of the brightest up-and-coming minds to his lab. "He’s developed those relationships, worked closely with people in summer programs, and that’s made it easier to bring in top-notch people."

Arthur says that one of his keys to being a good interviewer and creating a workplace where people want to join is avoiding unsolvable problems during those early conversations. "I can come up with 10 questions you’ll never answer in an hour, but that’s not interesting to me," he said. "I want to know about your body of research over five years, when you had the luxury of time. Have you come up with something interesting? Or are you just doing derivative work? If I give you a week, can you come up with something great?"

That focus on continuous innovation also helps keep the lab moving forward with a close-knit team in a competitive market. "I know that if I keep working with him, I’m going to grow," Tarek said.

AI and asphalt

In the Machine Learning Lab, of course, cars are involved. Talk to Arthur about a future with autonomous vehicles and he gets wistful. "My daily driver is a 2007 BMW Z4 M Coupe, and I like it because it has almost zero aids – there is little between you and what the car is doing on the road," he said. "But it’s also no fun driving in rush-hour Dallas traffic."

His dream is to own a 1987 RUF CTR, a limited-production modified Porsche 911 that he first saw as a teenager. Maybe one day in the future, when there are many autonomous cars on the road, he’ll be driving one with a smile on his face, a happy human in a world of intelligent machines.