Increasing student grades in large online subjects: Combining tutorial support with technology

Three large first-year undergraduate subjects with 240-517 enrolled students were selected to participate in this pilot study. A meeting scheduling tool was embedded in the learning management system and thirty-minute, one-on-one tutorial sessions were available to students in the 2 weeks leading up to the due date of at least one large written task. Thirty one percent (31%) of enrolled students attended at least one appointment with a tutor. There was no difference in the average assessment mark that students obtained before the first tutorial was offered between those who attended a tutorial session for a later assessment item and those who did not. There was a significant increase in the average cumulative grade (10%, p<0.05) of students who attended a tutorial. The novel use of the calendar booking tool combined with online meeting technology provides a simple and convenient method to provide personalised feedback to a large cohort of students.

Improving Multi-agent Coordination by Learning to Estimate Contention

We present a multi-agent learning algorithm, ALMA-Learning, for efficient and fair allocations in large-scale systems. We circumvent the traditional pitfalls of multi-agent learning (e.g., the moving target problem, the curse of dimensionality, or the need for mutually consistent actions) by relying on the ALMA heuristic as a coordination mechanism for each stage game. ALMA-Learning is decentralized, observes only own action/reward pairs, requires no inter-agent communication, and achieves near-optimal (<5% loss) and fair coordination in a variety of synthetic scenarios and a real-world meeting scheduling problem. The lightweight nature and fast learning constitute ALMA-Learning ideal for on-device deployment.

Meeting Scheduling Application using Availability and Priority Attendance Heuristics

Meeting scheduling is a repetitive and time consuming task for many organizations. Emails and electronic calendars has been used to help a meeting host in this process. However, it does not automate the process of searching the optimal time slot. Manual scheduling may result in suboptimal schedule. Therefore, automation is needed for meeting scheduling problem. The purpose of this research is to propose an applied model consisting of both acquiring participants’ existing schedule, and searching for an optimal time slot. Previous studies groups the solution of meeting scheduling into either constraint satisfaction or heuristics approach. Heuristics is more appropriate for a dynamic environment. The heuristics-based model is designed to consider participant availability and participant prioritization. The more participants are available, the better the time is as a candidate for optimum schedule. In the proposed model, the availability of certain key person, experts, or host may carry more weight than normal participant. An Android based application is developed as a prove of concept of the proposed model. Google Calendar API is used in this model to acquire the existing schedule, then each time slot is assigned a score based on availability weighting. The time slot with the highest score is considered the optimal solution. Evaluation is done by simulating the scheduling part for various numbers of meetings and time slots. The result shows that the model is capable of searching the optimal meeting schedule in less than one second for each of the experiment.

Integrating Meeting and Individual Events Scheduling

Automated meeting scheduling is the task of reaching an agreement on a time slot to schedule a new meeting, taking into account the participants’ preferences over various aspects of the problem. Such a negotiation is commonly performed in a non-automated manner, that is, the users decide whether they can reschedule existing individual activities and, in some cases, already scheduled meetings in order to accommodate the new meeting request in a particular time slot, by inspecting their schedules. In this work, we take advantage of SelfPlanner, an automated system that employs greedy stochastic optimization algorithms to schedule individual activities under a rich model of preferences and constraints, and we extend that work to accommodate meetings. For each new meeting request, participants decide whether they can accommodate the meeting in a particular time slot by employing SelfPlanner’s underlying algorithms to automatically reschedule existing individual activities. Time slots are prioritized in terms of the number of users that need to reschedule existing activities. An agreement is reached as soon as all agents can schedule the meeting at a particular time slot, without anyone of them experiencing an overall utility loss, that is, taking into account also the utility gain from the meeting. This dynamic multi-agent meeting scheduling approach has been tested on a variety of test problems with very promising results.