Self-Taught: Software That Learns By Doing

Machine-learning techniques to create self-improving software are hitting the mainstream.

By Gary Anthes

February 6, 2006 12:00 PM ET

Computerworld - Attempts to create self-improving software date to the 1960s. But "machine learning," as it's often called, has remained mostly the province of academic researchers, with only a few niche applications in the commercial world, such as speech recognition and credit card fraud detection. Now, researchers say, better algorithms, more powerful computers and a few clever tricks will move it further into the mainstream.


Stanford professor Sebastian Thrun with "Stanley," the car that used machine-learning techniques to drive itself 132 miles across the desert.

And as the technology grows, so does the need for it. "In the past, someone would look at a problem, write some code, test it, improve it by hand, test it again and so on," says Sebastian Thrun, a computer science professor at Stanford University and the director of the Stanford Artificial Intelligence Laboratory. "The problem is, software is becoming larger and larger and less and less manageable. So there's a trend to make software that can adapt itself. This is a really big item for the future."
Thrun used several new machine-learning techniques in software that literally drove an autonomous car 132 miles across the desert to win a $2 million prize for Stanford in a recent contest put on by the Defense Advanced Research Projects Agency. The car learned road-surface characteristics as it went. And machine-learning techniques gave his team a productivity boost as well, Thrun says. "I could develop code in a day that would have taken me half a month to develop by hand," he says.
Computer scientist Tom Mitchell, director of the Center for Automated Learning and Discovery at Carnegie Mellon University, says machine learning is useful for the kinds of tasks that humans do easily -- speech and image recognition, for example -- but that they have trouble explaining explicitly in software rules. In machine-learning applications, software is "trained" on test cases devised and labeled by humans, scored so it knows what it got right and wrong, and then sent out to solve real-world cases.
Mitchell is testing the concept of having two classes of learning algorithms in essence train each other, so that together they can do better than either would alone. For example, one search algorithm classifies a Web page by considering the words on it. A second one looks at the words on the hyperlinks that point to the page. The two share clues about a page and express their confidence in their assessments.
Mitchell's experiments have shown that such "co-training" can reduce errors by more than a factor

Recommend Digg Twitter ShareThis

Email Print Send Feedback Reprints

Additional Resources

POLL RESULTS

Leading IT Pros Weigh In on Top IT Issues

Accelerate your knowledge of the IT world you inhabit by viewing the results of a series of polls taken by your IT peers. These polls of 100+ IT professionals each are available for full viewing. They cover key topics such as virtualization, processor performance, green IT, cloud computing and many others. Be a part of the buzz.

WHITE PAPER

For Best Results, Think Beyond the Box

Technology is complex. Keeping it running productively shouldn't be. To that end, you want to minimize the number of solutions needed in-house to simplify operations, maintenance, and support. Kodak offers a best-practices model. One company provides support for both scanner and software, for fast problem resolution without vendor finger-pointing. Download now!

WHITE PAPER

Accelerating the Path to Demand Intelligence with a Demand Signal Repository

Utilizing demand intelligence improves the precision of pricing, product assortments, channel/store placement, and promotion, which are all essential for sustainable revenue management performance. Learn more, download this free whitepaper today.

White Papers & Webcasts

What your IT equipment needs from a UPS
What your IT equipment needs from a UPS: The top five requirements that define "quality power" in the eyes of the power supplies...

Strategic ECM Webinar
Learn what new strategic business benefits can be realized through ECM!...

Should Your Email Live in the Cloud - A Comparative Cost Analysis
Does cloud-based email make sense for your company? This report helps you calculate your onsite email costs and compare them to cloud-based alternatives....

Managing And Protecting Your Ever Increasing Mobile Assets
Learn best practices for desktop and application virtualization, computer security, and computer life-cycle management....

Impact of the Dramatic Increase in Devices on the Cost to Support
This white paper describes the challenges that CIO's will face in coming years due to a dramatic increase in the number of devices...

5 Architecture Issues that Impact BES performance
This Live webinar will identify critical log file errors, performance counters, and configurations to pay close attention to when optimizing BES server performance....

Hidden Cash: Maximizing the Value of Surplus Technology in a Down Economy
In today's tightened economy, all major technology purchases are being carefully scrutinized to ensure that each new piece of hardware and software can...

Usability Is Everything
Learn what sets Workday's HR and Payroll solutions apart from the competition....

Your Network at Half the Price: Slash Network Hardware Costs With Pre-Owned Equipment
Pre-owned networking equipment is certainly less expensive than the new variety, but IT managers are often challenged to know when and how to...

The Value of Real SaaS at Workday
Cost savings, speed to value, and innovation brought to the enterprise by Workday's software-as-a-service solutions for HR and Payroll....

All White Papers | All Webcasts