Skip navigation
Help

Andrew Ng

warning: Creating default object from empty value in /var/www/vhosts/sayforward.com/subdomains/recorder/httpdocs/modules/taxonomy/taxonomy.pages.inc on line 33.

There seems to be something in the water at Stanford University that’s making faculty members leave their more-than-perfectly-good jobs and go teach online.

Coursera co-founders Andrew Ng and Daphne Koller

Stanford computer science professors Daphne Koller and Andrew Ng are on leave to launch Coursera, which will offer university classes for free online, in partnership with top schools.

Mountain View, Calif.-based Coursera is backed with $16 million in funding led by John Doerr at Kleiner Perkins and Scott Sandell at NEA. It has no immediate plans to charge for courses or make money in other ways.

Compared to Udacity, a similar start-up from former Stanford professor Sebastian Thrun that’s creating its own classes, Coursera helps support its university partners in creating their own courses, which are listed under each school’s brand.

Some might doubt that universities would want to share their prized content for free online with a start-up, but Coursera has already signed up Princeton, Stanford, the University of Michigan and the University of Pennsylvania as partners, with a set of classes launching April 23.

Coursera evolved in part out of the hugely popular Stanford classes that Ng and Thrun taught last fall on machine learning and artificial intelligence, respectively. Ng’s course had 104,000 people enrolled, with at least 46,000 completing at least one homework assignment. Of those, 23,000 completed a “substantial” amount of the class, and 13,000 received a “statement of accomplishment,” Ng said. He’ll be offering that class again starting next week.

Koller and Ng are particularly committed to developing pedagogy for this new medium, and have built their own course software and student forums. They describe their philosophy as similar to that of Salman Khan and the Khan Academy, where students are encouraged to take the time to master material at their own pace.

Coursera students help other students — in the fall, the median response time to a question asked on the class forum was 22 minutes — and the system will also learn from the students.

For instance, 2,000 of the 20,000 or so students in Ng’s online class had the exact same wrong solution on one problem set, he said. That’s an opportunity to recognize what’s happening and teach those students in that moment.

Plus, Koller and Ng have also conceived of an ambitious plan to grade humanities classes with thousands of students enrolled.

Coursera’s content is naturally heavy on computer science — where problem sets are fairly straightforward to grade — but it will also offer poetry, sociology and medical courses. These classes will be graded crowdsourcing style, with peer assessment and review. Figuring out how to grade masses of assignments on a subjective scale is a machine learning problem, Ng said.

Another ambitious venture-backed college-level online education start-up I recently covered is the Minerva Project, which is planning to launch its own mostly virtual elite university.

0
Your rating: None


Bay Area Vision Meeting: Unsupervised Feature Learning and Deep Learning

Bay Area Vision Meeting (more info below) Unsupervised Feature Learning and Deep Learning Presented by Andrew Ng March 7, 2011 ABSTRACT Despite machine learning's numerous successes, applying machine learning to a new problem usually means spending a long time hand-designing the input representation for that specific problem. This is true for applications in vision, audio, text/NLP, and other problems. To address this, researchers have recently developed "unsupervised feature learning" and "deep learning" algorithms that can automatically learn feature representations from unlabeled data, thus bypassing much of this time-consuming engineering. Building on such ideas as sparse coding and deep belief networks, these algorithms can exploit large amounts of unlabeled data (which is cheap and easy to obtain) to learn a good feature representation. These methods have also surpassed the previous state-of-the-art on a number of problems in vision, audio, and text. In this talk, I describe some of the key ideas behind unsupervised feature learning and deep learning, describe a few algorithms, and present case studies pertaining. The Bay Area Vision Meeting (BAVM) is an informal gathering (without a printed proceedings) of academic and industry researchers with interest in computer vision and related areas. The goal is to build community among vision researchers in the San Francisco Bay Area, however, visitors and travelers from afar are also encouraged to attend and present. New <b>...</b>
From:
GoogleTechTalks
Views:
4431

62
ratings
Time:
48:10
More in
Science & Technology

0
Your rating: None