Goto

Collaborating Authors

Efficient Online Learning via Randomized Rounding

Neural Information Processing Systems

Most online algorithms used in machine learning today are based on variants of mirror descent or follow-the-leader. In this paper, we present an online algorithm based on a completely different approach, which combines random playout'' and randomized rounding of loss subgradients. As an application of our approach, we provide the first computationally efficient online algorithm for collaborative filtering with trace-norm constrained matrices. As a second application, we solve an open question linking batch learning and transductive online learning. Papers published at the Neural Information Processing Systems Conference.


Efficient Second Order Online Learning by Sketching

Neural Information Processing Systems

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. Papers published at the Neural Information Processing Systems Conference.


A Boosting Framework on Grounds of Online Learning

Neural Information Processing Systems

By exploiting the duality between boosting and online learning, we present a boosting framework which proves to be extremely powerful thanks to employing the vast knowledge available in the online learning area. Using this framework, we develop various algorithms to address multiple practically and theoretically interesting questions including sparse boosting, smooth-distribution boosting, agnostic learning and, as a by-product, some generalization to double-projection online learning algorithms. Papers published at the Neural Information Processing Systems Conference.


Universities' move online 'must be done the right way'

BBC News

Harry Ashworth should be in the final term of his first year at Oxford University, studying music. Instead he is stuck at his parents' house in south London hunched over a laptop, listening to lectures via Zoom. He doesn't feel that the sudden and dramatic change in circumstances has affected his learning too much, but he is missing some aspects of university life. "I am in a jazz orchestra and that isn't really happening now. And I would have been playing at the summer balls, so there are social events that I've missed."


4 ways to stimulate online learner engagement NEO BLOG

#artificialintelligence

Online learning is a highly engaging format. Most students are immediately stimulated by an asynchronous format that places them at the centre of their learning journey. The range of multimedia resources, variety of interactive platforms, and the thrill of being able to manage their time and other resources boosts curiosity and motivation. For most students the online learning experience is a freeing leap into a new way of learning. However, as an online instructor you are no doubt aware of those students that are struggling to make the transition.