We consider a novel “online semi-supervised learning” setting where (mostly unlabeled) data arrives sequentially in large volume, and it is impractical to store it all before learning. We propose an online manifold regularization algorithm. It differs from standard online learning in that it learns even when the input point is unlabeled. Our algorithm is based on convex programming in kernel sp...

The traditional co-training algorithm, which needs a great number of unlabeled examples in advance and then trains classifiers by iterative learning approach, is not suitable for online learning of classifiers. To overcome this barrier, we propose a novel semi-supervised learning algorithm, called MAPACo-Training, by combining the co-training with the principle of Maximum A Posteriori adaptatio...

In this paper we develop a mechanism for critic-based learning in continuous state and action spaces. Our approach is based on the Motoric Map model [RMS90], by which we wish to overcome the restrictions of traditional Reinforcement Learning methods concerning continuous spaces. Covariance Learning is introduced as algorithm to determine the best possible action for a given state using the crit...

We describe a novel framework for the design and analysis of online learning algorithms. Our framework is based on a new perspective on relative mistake bounds by viewing the number of mistakes of an online learning algorithm as a lower bound for an optimization problem. This interpretation of a mistake bound draws a connection between online learning and optimization through the theory of dual...

In this paper a reinforcement learning methodology for automatic online algorithm selection is introduced and empirically tested. It is applicable to automatic algorithm selection methods that predict the performance of each available algorithm and then pick the best one. The experiments confirm the usefulness of the methodology: using online data results in better performance. As in many onlin...

We propose a kernel-based online semi-supervised algorithm that is applicable for large scale learning tasks. In particular, we use a multi-view learning framework and a co-agreement strategy to take into account unlabelled data and to improve classification performance of the algorithm. Unlike the standard online methods our algorithm is naturally applicable to many real-world situations where...

Introduction: Postgraduate medical education involves the use ofonline-learning tools. However, there is a paucity of data on theuse of online-learning among doctors who are in their 1st and 2ndyears of professional work after graduating from medical school(also known as Foundation doctors). Our aim was to explore theuse of online-learning among Foundation doctors.Methods: A cross-sectional stu...