Virtual learning programs could soon mark the end of test-taking. Researchers from Stanford University and Google in California have developed a new algorithm that aims to understand students' individual learning abilities. The system can even predict whether the students will get questions right or wrong in a given exercise, and as development progresses, it may be also able to identify why they've chosen that way. Virtual learning programs could soon mark the end of test-taking. Researchers from Stanford University and Google in California have developed a new algorithm that can understand students' individual learning abilities.

To some degree, that's inevitable: Many people who enroll in MOOCs may have no interest in doing homework, but simply plan to listen to video lectures in their spare time. Others, however, may begin courses with the firm intention of completing them but get derailed by life's other demands. Identifying those people before they drop out and providing them with extra help could make their MOOC participation much more productive. The problem is that you don't know who's actually dropped out -- or, in MOOC parlance, "stopped out" -- until the MOOC has been completed. One missed deadline does not a stopout make; but after the second or third missed deadline, it may be too late for an intervention to do any good.

MIT President L. Rafael Reif announced today a significant expansion of the Institute's programs in learning research and online and digital education -- from pre-kindergarten through residential higher education and lifelong learning -- that fulfills a number of recommendations made in 2014 by the Institute-Wide Task Force on the Future of MIT Education. Most notably, Reif announced the creation of the MIT Integrated Learning Initiative (MITili), to be led by Professor John Gabrieli, and a new effort to increase MIT's ability to improve science, technology, engineering, and mathematics (STEM) learning by students from pre-kindergarten through high school (pK-12), to be led by Professor Angela Belcher. The announcement also included a program to support faculty innovations in MIT residential education and new work to enhance MIT's continuing education programs. In keeping with the high priority of these new efforts and of the entire field of digital learning, Professor Sanjay Sarma, now dean of digital learning, will oversee them in the newly created position of vice president for open learning, reporting directly to Reif. Chancellor Cynthia Barnhart, who will share responsibility with Sarma for several aspects of this work, predicts that the programs announced today will have "far-reaching and tremendous implications for education -- for MIT students as well as for students not at MIT."

Greater compute power and power efficiency has made deep learning algorithms ubiquitous in our world. Deep learning has found its way into self driving cars, convenience stores and hospitals. Yet the fight for top talent in the space remains fierce and is a bottleneck for reaching new industries and solving tough challenges. To complement Udacity's previous AI courses, the online education startup is partnering with YouTube star Siraj Raval for a new deep learning nanodegree foundation program that will be co-taught with Udacity's Mat Leonard. Foundation Programs are going to be a major focus for Udacity in the coming year.

What don't you know that you need to know, and how do you know you don't know? Imagine putting that into a search engine, expecting a coherent answer. The answers we seek are at the core of discovery and learning, motivated by necessity and pleasure, by job displacement, or leadership uncertainty in a complex and fast changing world. The process of learning requires a detailed knowledge of ourselves and of our world, whether a human or a machine is tasked to help. We can fill gaps in our skills and knowledge through web searches, discussion with experts and like-minded people, reading books, working through an education curriculum, learning online with video tutorials, and absorbing a vast amount of content flowing through online news channels and aggregators.

You've probably heard it before: "a bad craftsman blames his tools." But when is the last time you saw someone building a house with a hammer, a hand saw and some 2x4s? When you build a house, you need the right tools and materials to build a house. When you build a skills, there are a different set of tools and materials. The basic ingredients for learning are neurons and myelin. Each time you fire a set of neurons while learning, they get wrapped in another thin layer of myelin, which is like insulation on an electric cord.

Greetings from CES, the annual consumer electronics extravaganza in Las Vegas that began as an opportunity for retailers to see what gadgets they'd be buying to sell to their customers in the coming year and has evolved into a see-and-be-seen event for the technology industry's elite. I led a conversation with Andrew Ng, the leader of Chinese Internet giant Baidu's artificial intelligence research unit. Ng is a Stanford professor who helped start Google Brain and co-founded online education firm Coursera. He is one of the world's foremost experts on AI who speaks clearly about the complicated topic in terms businesspeople can relate to. Get Data Sheet, Fortune's technology newsletter, where this essay originated. AI will become so important and is so confusing to most businesspeople that only an expert can help business leaders make sense of it.

This is a follow up to an article Per wrote last year, Machine Learning in a Week, on how he kickstarted his way into machine learning (ml) by devoting five days to the subject. Follow him on Medium and check out his archive. My interest in ml stems back to 2014 when I started reading articles about it on Hacker News. I simply found the idea of teaching machines stuff by looking at data appealing. At the time I wasn't even a professional developer, but a hobby coder who'd done a couple of small projects. So I began watching the first few chapters of Udacity's Supervised Learning course, while also reading all articles I came across on the subject.

We propose Sketched Online Newton (SON), an online second order learning algorithm that enjoys substantially improved regret guarantees for ill-conditioned data. SON is an enhanced version of the Online Newton Step, which, via sketching techniques enjoys a running time linear in the dimension and sketch size. We further develop sparse forms of the sketching methods (such as Oja's rule), making the computation linear in the sparsity of features. Together, the algorithm eliminates all computational obstacles in previous second order online learning approaches.

In the recent years, a number of parameter-free algorithms have been developed for online linear optimization over Hilbert spaces and for learning with expert advice. These algorithms achieve optimal regret bounds that depend on the unknown competitors, without having to tune the learning rates with oracle choices. We present a new intuitive framework to design parameter-free algorithms for both online linear optimization over Hilbert spaces and for learning with expert advice, based on reductions to betting on outcomes of adversarial coins. We instantiate it using a betting algorithm based on the Krichevsky-Trofimov estimator. The resulting algorithms are simple, with no parameters to be tuned, and they improve or match previous results in terms of regret guarantee and per-round complexity.