February 2011

Host: Robyn McNamara (Faculty of IT)

Titile: Computer Programming Assessment for Computer Science Courses

Abstract: Learning to program is hard.  Assessing students' programming ability seems to be even harder.  Studies from around the world have shown that many students leave their introductory programming courses unable to construct a relatively simple program from scratch.  We need to be sure that students' results genuinely reflect their capabilities, not only to determine whether they are ready to proceed to further study, but to understand how effective our teaching is and how it might be improved.  In order to construct valid assessment, we are first going to have to understand how students' skills develop as they proceed from novice to master.  In this talk I will describe a framework for classifying students' developmental stages, discuss its ramifications for the validity of the assessment we currently use, and present some ideas for better assessment.

March 2011

Host: Upuli Gunasinghe

Title: 'A cognition inspired approach to capturing data sequences'

Abstract: From engineering to health and from finance to scientific research, sequences naturally arise in many domains. Although many attempts have been taken to address the sequence learning problem a lot more can be done to make sequence learning more effective. Humans have the amazing ability to capture, understand and use sequences in everyday cognitive tasks such as language, speech, motor control, vision, problem solving etc. Therefore this research takes inspiration from human intelligence and ability in sequence learning to come up with a solution.  It looks at one of the most widely used biologically inspired learning rules, the Hebb Rule, to develop a model to capture sequences. In order to address problems in continuous input spaces, it is expected to integrate the proposed model with the unsupervised clustering techniques SOM and GSOM. Inspired by the Memory Prediction Framework, the final goal of the research is to build an architecture, which can be used to continuously accumulate knowledge in a human like manner. This presentation is based on the work carried out for Upuli's PhD confirmation.

Host: David Abramson

Title: 'e-Science: Are we there yet?'

Abstract: e-Science involves the application of advanced computational methods to other areas of science and technology. It It has attracted a good deal of support over the past 10 years, and numerous groups have developed new techniques and software prototypes. Importantly, e-Science requires advanced in both computer science and the application area, making it an ideal driver for computer science research.

In this talk, I will explore whether any of this work is actually making a difference. I will discuss our own projects work at the Monash e-Science and Grid Engineering (MeSsAGE) Lab, a computer science research laboratory devoted to new software development techniques that support e-Science applications. I will show how high throughput (aka parallel) scientific workflows have not only contributed to the state of the art in computer science, but are being adopted in research labs at Monash and internationally. In particular, I will highlight case studies in the medical imaging, chemistry and cardiac science.

April 2011

Host: Mohammed Al-Naeem

Title: 'A Rapid Target Detection Scheme for WSNs'

Abstract: Wireless sensor networks are an exciting and promising domain of highly networked systems of low-power wireless nodes. They also have a very limited capability of CPU processing and memory. These may be deployed as large integrated networks for macroscopic viewing of the environment. A new aim of sensor nodes in a wireless sensor network is to provide a birds-eye of the observed area by interacting and exchanging information. This research is focusing on resolving target issues in WSNs, which require developing a rapid scheme that is capable to rapidly detect targets when they enter the field of interest. Target detection process must involve recognising these targets to ensure that they are, in fact, the targets’ objects. Also, the proposed scheme should be capable of tracking the target along the field of interest, which will lead to being able to identify the target direction and predict its final destination in the field of interest.

Host: Marsha Minchenko

Title: 'Searching for certain Cayley integral graphs'

Abstract: Cayley graphs were designed to help understand the abstract nature of groups. Being very symmetric, they proved to have many more uses than this.  Given a group, G, and a generating set of that group, S; the vertices of the cayley graph Γ = Γ(G,S) are the elements of the group, G, and the edges, given any   are all the ordered pairs (g,gs).  A graph is integral if all its eigenvalues, with respect to the adjacency matrix of the graph, are integers. In this talk I'd like outline a specific set of Cayley Integral graphs and then explain why and how I would bother searching for them.

Tuesday, 24 May 2011, 1-2pm (Room 135/26)

Host: Yanir Seroussi

Title: (1) Personalised Rating Prediction for New Users Using Latent Factor Models

         

(2) Authorship Attribution with Latent Dirichlet Allocation

Abstract:

In this talk, I will present two papers I wrote with Ingrid Zukerman and Fabian Bohnert, in preparation for my June conference trip. I will be presenting Paper 1 at Hypertext 2011 (http://www.ht2011.org) and Paper 2 at CoNLL 2011 (http://www.clips.ua.ac.be/conll). See the papers' abstracts below.

(1) In recent years, personalised recommendations have gained importance in helping users deal with the abundance of information available online. Personalised recommendations are often based on rating predictions, and thus accurate rating prediction is essential for the generation of useful recommendations. Recently, rating prediction algorithms that are based on matrix factorisation have become increasingly popular, due to their high accuracy and scalability. However, these algorithms still deliver inaccurate rating predictions for new users, who submitted only a few ratings.

In this paper, we address the new user problem by introducing several extensions to the basic matrix factorisation algorithm, which take user attributes into account when generating rating predictions. We consider both demographic attributes, explicitly supplied by users, and attributes inferred from user-generated texts. Our results show that employing our text-based user attributes yields personalised rating predictions that are more accurate than our baselines, while not requiring users to explicitly supply any information about themselves and their preferences.

(2) The problem of authorship attribution – attributing texts to their original authors – has been an active research area since the end of the 19th century, attracting increased interest in the last decade. Most of the work on authorship attribution focuses on scenarios with only a few candidate authors, but recently considered cases with tens to thousands of candidate authors were found to be much more challenging. In this paper, we propose ways of employing Latent Dirichlet Allocation in authorship attribution. We show that our approach yields state-of-the-art performance for both a few and many candidate authors, in cases where these authors wrote enough texts to be modelled effectively.

Host: Chang Joo Yun (River)

Title: "Intelligent Multi-Criteria Decision Making Models for Supply Chain Management"

Abstract: 

This study develops a methodological framework, which has two steps to select the supplier group and choose the suitable suppliers in the selected group based on the strategic and operational criteria, using the customer order dependent weighting method and the MCDM model. An empirical study on an auto parts manufacturing company is conducted to demonstrate the effectiveness of the model. The strategic criteria can be identified by considering the relationship between the auto parts company, its customers and suppliers, the strategic plan of the company for supplier development, the characteristics of Daegu, and the breakeven criteria. A customer order dependent weighting method is developed to determine the weights of the the supplier selection criteria. This is achieved by modeling the relationship between the customer order factors and the supplier selection criteria, using a knowledge base with if-then rules. With the elicited selection criteria weights, an MCDM model is used to select among suitable suppliers for a given customer order. More effective decision making can be achieved by identifying the relationship between the customer order factors and the supplier selection criteria. The structured two-step approach will enable the supplier development and the customer-supplier relationship enhancement of the auto parts company.

Host: Dhananjay Thiruvady (DJ)

Title: "Car Sequencing with Constraint-Based ACO"

Abstract:

Hybrid methods for solving combinatorial optimization problems have become increasingly popular recently. This study is concerned with hybrids of ant colony optimization and constraint programming which are typically useful for problems with hard constraints. However, the original algorithm suffered from large CPU time requirements. It was shown that this integration can be made efficient via a further hybridization with beam search resulting in CP-Beam-ACO. The original work suggested this in the context of job scheduling. We show here that this algorithm type is also effective on another problem class, namely car sequencing. We consider an optimization version, where we aim to optimize the utilization rates across the sequence. Car sequencing is notoriously difficult problem, because it is difficult to obtain good bounds via relaxations. We show that stochastic sampling provides superior results to well known lower bounds for this problem when combined with CP-Beam-ACO.