Online discovery of similarity mappings

We consider the problem of choosing, sequentially, a map which assigns elements of a set A to a few elements of a set B. On each round, the algorithm suffers some cost associated with the chosen assignment, and the goal is to minimize the cumulative loss of these choices relative to the best map on the entire sequence. Even though the offline problem of finding the best map is provably hard, we show that there is an equivalent online approximation algorithm, Randomized Map Prediction (RMP), that is efficient and performs nearly as well. While drawing upon results from the "Online Prediction with Expert Advice" setting, we show how RMP can be utilized as an online approach to several standard batch problems. We apply RMP to online clustering as well as online feature selection and, surprisingly, RMP often outperforms the standard batch algorithms on these problems.

Implicit online learning

Online learning algorithms have recently risen to prominence due to their strong theoretical guarantees and an increasing number of practical applications for large-scale data analysis problems. In this paper, we analyze a class of online learning algorithms based on fixed potentials and nonlinearized losses, which yields algorithms with implicit update rules. We show how to efficiently compute these updates, and we prove regret bounds for the algorithms. We apply our formulation to several special cases where our approach has benefits over existing online learning methods. In particular, we provide improved algorithms and bounds for the online metric learning problem, and show improved robustness for online linear prediction problems. Results over a variety of data sets demonstrate the advantages of our framework.

Optimal strategies and minimax lower bounds for online convex games

A number of learning problems can be cast as an Online Convex Game: on each round, a learner makes a prediction x from a convex set, the environment plays a loss function f, and the learner’s long-term goal is to minimize regret. Algorithms have been proposed by Zinkevich, when f is assumed to be convex, and Hazan et al., when f is assumed to be strongly convex, that have provably low regret. We consider these two settings and analyze such games from a minimax perspective, proving minimax strategies and lower bounds in each case. These results prove that the existing algorithms are essentially optimal.

Matrix regularization techniques for online multitask learning

In this paper we examine the problem of prediction with expert advice in a setup where the learner is presented with a sequence of examples coming from different tasks. In order for the learner to be able to benefit from performing multiple tasks simultaneously, we make assumptions of task relatedness by constraining the comparator to use a lesser number of best experts than the number of tasks. We show how this corresponds naturally to learning under spectral or structural matrix constraints, and propose regularization techniques to enforce the constraints. The regularization techniques proposed here are interesting in their own right and multitask learning is just one application for the ideas. A theoretical analysis of one such regularizer is performed, and a regret bound that shows benefits of this setup is reported.

Closing the gap between bandit and full-information online optimization : high-probability regret bound

We demonstrate a modification of the algorithm of Dani et al for the online linear optimization problem in the bandit setting, which allows us to achieve an O( \sqrt{T ln T} ) regret bound in high probability against an adaptive adversary, as opposed to the in expectation result against an oblivious adversary of Dani et al. We obtain the same dependence on the dimension as that exhibited by Dani et al. The results of this paper rest firmly on those of Dani et al and the remarkable technique of Auer et al for obtaining high-probability bounds via optimistic estimates. This paper answers an open question: it eliminates the gap between the high-probability bounds obtained in the full-information vs bandit settings.

Adaptive online gradient descent

We study the rates of growth of the regret in online convex optimization. First, we show that a simple extension of the algorithm of Hazan et al eliminates the need for a priori knowledge of the lower bound on the second derivatives of the observed functions. We then provide an algorithm, Adaptive Online Gradient Descent, which interpolates between the results of Zinkevich for linear functions and of Hazan et al for strongly convex functions, achieving intermediate rates between [square root T] and [log T]. Furthermore, we show strong optimality of the algorithm. Finally, we provide an extension of our results to general norms.

Doorway to research : Issues in the development of an interactive website for international graduate students

This paper describes results of a study evaluating the content, functionality and design features of an innovative online website called the Doorway to Research (http://rsc.acid.net.au/Main.aspx) , which was developed to support international graduate students studying at universities in Australia. First, the key features of the website are described. Second, the result of a pilot study involving 12 students and faculty members who tested key aspects of the design, content and functionality of the website and provided written and oral feedback base on task-based questions and focus group discussions are explored. Finally, recommendations for future development are presented. Results of the study indicate general student satisfaction with the website and its design, content and functionality, with specific areas identified for further development.

Ensuring student success : the role of support services in improving the quality of the student learning experience

This paper is based on the premise that universities have an obligation to provide adequate student support services, such as learning assistance (that is, assistance with academic writing and other study skills) and that in order to be effective such services must be responsive to the wider policy and social implications of student attrition and retention. The paper outlines briefly some of the factors that have influenced the development of learning assistance practices in Australia and America. This is followed by an account of experiences at one Australian metropolitan university where learning assistance service provision shifted from a decentralised, faculty-based model to a centralised model of service delivery. This shift was in response to concerns about lack of quality and consistency in a support model dependent upon faculty resources yet a follow up study identified other problems in the centralised delivery of learning assistance services. These problems, clustered under the heading contextualised versus decontextualised learning assistance, include the relevance of generic learning assistance services to students struggling with specific course related demands; the apparent tensions between challenging students and assisting students at risk of failure; and variations in the level of collaboration between learning advisers and academic staff in supporting students in the learning environment. These problems are analysed using the theoretical modelling derived from the tools made available through cultural historical activity theory and expansive visibilisation (Engeström & Miettinen, 1999).

The Chronic Heart-Failure Assistance by Telephone (CHAT) Study: Assessment of Telephone Support for Vulnerable Patients with Chronic Disease

Aim: To determine whether telephone support using an evidence-based protocol for chronic heart failure (CHF) management will improve patient outcomes and will reduce hospital readmission rates in patients without access to hospital-based management programs. Methods: The rationale and protocol for a cluster-design randomised controlled trial (RCT) of a semi-automated telephone intervention for the management of CHF, the Chronic Heart-failure Assistance by Telephone (CHAT) Study is described. Care is coordinated by trained cardiac nurses located in Heartline, the national call center of the National Heart Foundation of Australia in partnership with patients’ general practitioners (GPs). Conclusions: The CHAT Study model represents a potentially cost-effective and accessible model for the Australian health system in caring for CHF patients in rural and remote areas. The system of care could also be readily adapted for a range of chronic diseases and health systems. Key words: chronic disease management; chronic heart failure; integrated health care systems; nursing care, rural health services; telemedicine; telenursing

Nursing Sans Frontieres: A Three Year Case Study of Multi-State Registration of Support Nursing Practice Using Information Technology

Objective: To highlight the registration issues for nurses who wish to practice nationally, particularly those practicing within the telehealth sector. Design: As part of a national clinical research study, applications were made to every state and territory for mutual recognition of nursing registration and fee waiver for telenursing cross boarder practice for a period of three years. These processes are described using a case study approach. Outcome: The aim of this case study was to achieve registration in every state and territory of Australia without paying multiple fees by using mutual recognition provisions and the cross-border fee waiver policy of the nurse regulatory authorities in order to practice telenursing. Results: Mutual recognition and fee waiver for cross-border practice was granted unconditionally in two states: Victoria (Vic) and Tasmania (Tas), and one territory: the Northern Territory (NT). The remainder of the Australian states and territories would only grant temporary registration for the period of the project or not at all, due to policy restrictions or nurse regulatory authority (NRA) Board decisions. As a consequence of gaining fee waiver the annual cost of registration was a maximum of $145 per annum as opposed to the potential $959 for initial registration and $625 for annual renewal. Conclusions: Having eight individual nurses Acts and NRAs for a population of 265,000 nurses would clearly indicate a case for over regulation in this country. The structure of regulation of nursing in Australia is a barrier to the changing and evolving role of nurses in the 21st century and a significant factor when considering workforce planning.