School of Computer Science and Software Engineering

Seminars 2011

School Research Seminars Presented in 2011

  1. Construction of polynomial-size optical priority queues using linear switches and fiber delay lines
  2. Tableaux for dense time temporal logic
  3. The identification of like-minded communities based on common interest in online social networks
  4. Automated prioritization of functional test cases for domain specific web applications
  5. Prediction market approaches to communication and coordintation strategies
  6. Safe information sharing among cheating partners - and an overview of some of my further research topics
  7. Explanatory visual debugging to assist novice C programmers
  8. A new approach for test case prioritization for regression testing of large software systems
  9. PCA for improving the performance of XCSR in classification of high dimensional problems
  10. SenPerfl: distributed passive performance monitoring of wireless sensor networks
  11. An evaluation of electronic individual peer assessment in an introductory programming
  12. Approximated nearest points for image set classification
  13. Robust and efficient encoding for multispectral palmprint recognition
  14. Novel bioreactor systems (Zetos/Chondros) for ex-vivo assessment of mechanical behaviour of bone and cartilage
  15. Retaining both privacy and utility in published network traces
  16. Fast and robust human recognition using 2D and 3D ear and face images

Construction of Polynomial-Size Optical Priority Queues Using Linear Switches nnd Fiber Delay Lines

  • Professor Amitava Datta - CSSE
  • 11am Friday 9th December
Abstract

One of the main problems in all-optical packet switched networks is the buffering of packets. A popular solution for buffering packets is to use a set of fiber delay lines attached to a switch. A priority queue is one of the most general buffering schemes that allows the packet with the highest priority to depart the buffer on a departure request and dropping of the least priority packet if a new packet arrives when the buffer is full.

We present a recursive algorithm for constructing optical priority queues of polynomial size from a switch with linear number of inputs/outputs and fiber delay lines. The best known lower bound allows the construction of a priority queue of size 2^M using a switch of size O(M). However, the best known upper bound constructs a priority queue of size O(M^3) using a switch of size O(M). We show that it is possible to construct a priority queue of size O(M^c) for any constant c using a switch of size O(M).
About the Presenter

Amitava Datta is a professor at the School of Computer Science and Software Engineering at UWA. His main research interests are in parallel and distributed computing, wireless networks, bioinformatics and (more recently) social networking. Amitava is a staff member in the school since 1998.

Back to top


Tableaux for Dense Time Temporal Logic

  • Ji Bian - PhD
  • 11am Friday 2nd December
Abstract

Practical reasoning aids for dense-time temporal logics are not at all common despite a range of potential applications from verification of concurrent systems to AI. There have been recent suggestions that the temporal mosaic idea can provide implementable tableau-style decision procedures for various linear time temporal logics beyond the standard discrete natural numbers model of time. In this presentation we report on investigations into the detailed algorithms needed to implement in this way a complete tableau reasoning procedure for the temporal logic of until and since over general linear time.

About the presenter

Ji Bian, Phd student of UWA in Temporal Logic group. The research area is tableaux for linear dense time. His supervisors are Mark Reynolds and Tim French.

Back to top


The Identification of Like-minded Communities Based on Common Interest in Online Social Networks

  • Kwan Hui Lim - MSc
  • 11am Friday 25th November
Abstract

Two of the most important research problems related to online social networks are the community detection in large social networks and efficient identification of interests among these communities. Current approaches for identifying interests involve first detecting all communities, followed by determining the interests of these communities. These approaches involve a lengthy and intensive process of detecting communities for the entire social network, which is growing daily. Furthermore, many of the detected communities may not share the interest we are looking for. Our research aims to solve this problem by proposing a community detection method for online social networks based on common interest. This method does not aim to detect all communities but detect communities that are both more cohesive and connected while comprising only of individuals sharing common interests.

Back to top


Automated Prioritization of Functional Test Cases for Domain Specific Web Applications

  • Deepak Garg - PhD
  • 11am Friday 18th November
Abstract

A test case prioritization technique schedules test cases for regression testing in an order that helps in achieving some specific goals. Current testing of complex web applications requires manual prioritization of test cases. This is a very time consuming task. In our research, we have proposed several new prioritization strategies for web applications and examined whether these prioritization strategies improve the rate of fault detection for web applications. We will also propose a new automated test suite prioritization approach for web applications that will automatically reschedule test cases and detect faults early in test suite execution.

Back to top


Prediction Market Approaches to Communication and Coordination Strategies

  • Aidan Morgan - PhD
  • 11am Friday 21st October
Abstract

This seminar will present experiences in participating in the UWA led WAMbot entry into the Multiple Autonomous Ground Vehicle Integration Challenge (MAGIC2010). The team produced a vehicle which successfully integrated hardware (sensors, vehicle platform) and software (sensing, planning, team coordination) into a cohesive and implementing communication and coordination strategies for a group of vehicles to work as a collaborative team for a number of tasks. In particular, MAGIC focused on minimising the human interaction in controlling a fleet of heterogeneous ground robots whilst completing a challenge with three main constraints: - Exploring and mapping the challenge area to an accuracy of +/- 0.5m; - Detecting, locating, classifying all objects of interest and recognising and neutralising all hostile objects - Completing within a time limit of 210 minutes. Secondly, the seminar will discuss the findings of a recent literature review. The research focuses on investigating the use of a special form of a Market Economy known as a Prediction Market and how it can be applied to the coordination and control of a fleet of unmanned vehicles, specifically concentrating on exploration and surveillance tasks. This seminar will explain what a Prediction Markets is, how they work and the advantages of using them for this problem.

Back to top


"Safe Information Sharing Among Cheating Partners - and an Overview of Some of My Further Research Topics"

  • Prof. Dr. Stefan Böttcher  - University of Paderborn Germany
  • 10.30am Monday 12th September
Abstract

The problem of safe information sharing is described using an example of two competing companies that want to find their common customers without telling the other company about all their customers. A further challenge is that we assume that the companies may use any technique to cheat, e.g. invent customers, suppress messages, and even stop the information sharing protocol at any point in time. We are nevertheless looking for a fair protocol that does not uncover information to one party only, even in all cases where one party cheats and the other party is honest. Beyond this key part, the talk gives an overview of some of my further research topics.

About the presenter

Dr. Stefan Böttcher is a Professor of Computer Science at the University of Paderborn Germany. He has published more than 100 refereed papers and has worked for or cooperated with more than 30 companies. His main research interests are algorithms on compressed XML and text data, theorem proving, access control, security, privacy, query processing, atomicity and synchronization of transactions, caching, and mobile ad-hoc networks.

Back to top


Explanatory Visual Debugging to Assist Novice C Programmers

  • Matthew Heinsen Egan - PhD
  • 11am Friday 9th September
Abstract

The C programming language has been widely used for decades, can be used to program a vast range of systems, and has influenced numerous other programming languages. C's use as an introductory programming language has waned, but it maintains a strong presence in other courses, primarily operating systems and networking. Many students have difficulty learning C, even when transitioning from other programming languages. In particular, C is often the first time that students are exposed to pointers, which are a Threshold Concept in Computer Science. Students may understand the concept of pointers, but have difficulty understanding their implementation and run-time behaviour. Program Visualization consists of the creation of graphical representations of information about a program (e.g. the program's source or run-time memory). It has proven effective in assisting students to understand the run-time behaviour of programs, and to debug their own programs. Unfortunately, novice-focused tools for C are typically outdated and unmaintained. Our project will focus on the design and development of novice-focused systems for debugging and understanding C programs, using modern debugging techniques and explanatory program visualization. This seminar will consider our motivations for focusing on the C programming language, previous novice-focused tools for C, the merits of novice-focused program visualization and debugging, and the specific goals of our project.

Back to top


A New Approach for Test Case Prioritization for Regression Testing of Large Software Systems

  • Deepak Garg – PhD
  • 11am Thursday 7th July
Abstract

Regression testing is applied to modified programs to identify whether the modified code behaves as intended. Computer programs like web applications or desktop applications are increasing in size due to the introduction of new features and functionalities. The need to perform regression testing becomes extremely important, as the changes in one function impact other functions. Sometimes, it becomes difficult to test the impact of these new /modified functions on existing tested functions as the execution of test cases for whole programs takes a long time. To reduce the execution time of test cases, we will suggest a new two-level algorithmic approach. This approach will first select the test cases for functionalities that are impacted by the changed functionality. We will detect the functions impacted by modified functions using dependency analysis. The dependency analysis will be performed using call graphs. Then we will prioritize the test cases by applying new/existing prioritization techniques. This new approach will be first investigated for procedural languages then we will proceed to object-oriented languages.

Back to top


PCA for Improving the Performance of XCSR in Classification of High-dimensional Problems

  • Mohammad Behdad - PhD
  • 12pm Monday 4th July
  • Weatherburn Lecture Theatre in Mathematics
Abstract

XCSR is an accuracy-based learning classifier system (LCS) which can handle classification problems with real-value features. However, as the number of features increases, a high classification accuracy comes at the cost of more resources: larger population sizes and longer computational running times. In this research, we present a PCA-enhanced LCS, which uses principal component analysis (PCA) as a preprocessing step for XCSR, and examine how it performs on complex multi-dimensional real-world data. The experiments show that this technique, in addition to significantly reducing the computational resources and time requirements of XCSR, maintains its high accuracy and even occasionally improves it. In addition to that, it reduces the required population size needed by XCSR.

Back to top


Senperfi: Distributed Passive Performance Monitoring of Wireless Sensor Networks

  • Thomas Buehring - University of the Federal Armed Forces, Munich, Germany
  • 11am Tuesday 21st June
Abstract

Network management and performance monitoring of Wireless Sensor Networks (WSNs) is an important problem because of the growing number of real-world applications of WSNs. In this seminar, I will present the results of my Master`s Thesis project registered at the University of the Federal Armed Forces at Munich, Germany. It studies an application of distributed passive monitoring to performance analysis of WSNs. I will give examples of scenarios and WSN platforms in order to derive specific requirements for a distributed passive monitoring solution. I will give an overview of applicable WSN performance metrics and discuss selected problems of the passive monitoring approach. I will also describe how the performance measurement approach can be embedded in a framework and talk about a case study of a specific WSN platform, which is being developed in a cooperation of the University of Western Australia and the University of Applied Sciences in Mannheim, Germany.

About the Presenter

Research student from the University of the Federal Armed Forces, Munich, Germany.

Back to top


An Evaluation of Electronic Individual Peer Assessment in an Introductory Programming

  • Dr Michael de Raadt - University of Southern Queensland
  • 11am Friday 17th June
Abstract

Peer learning is a powerful pedagogical practice delivering improved outcomes over conventional teacher-student interactions while offering marking relief to instructors. Peer-review enables learning by evaluating the work of others.

A system has been developed that facilitates anonymous review and delivers prompt feedback from multiple sources. This study is an evaluation of the use of this system in an introductory programming course. Use of the system is examined and attitudes of novice programmers towards the use of peer-review are compared to students from other disciplines raising a number of interesting issues. Recommendations are made to introductory programming instructors who may be considering peer-review in assignments

About the Presenter

Dr Michael de Raadt BSci(Hons), PhD Senior Lecturer (Computing), Computing Coordinator Dept. Mathematics and Computing, Faculty of Sciences Principal Advisor (Teaching and Learning) Division of ICT Services Development Manager, Moodle HQ University of Southern Queensland
Toowoomba, QLD, 4350, Australia

Back to top


Approximated Nearest Points for Image Set Classification

  • Yiqun Hu - PhD
  • 11am Friday 27th May
Abstract

Classification based on image sets has recently attracted great research interest as it holds more promise than single image based classification. In image set classification, each class is represented by one or more image sets and a query image set is assigned the label of the gallery set that is the nearest to it using some distance criterion. For the specific case of human faces, each set comprises a different number of facial images under arbitrary poses, illumination conditions and expressions. In this seminar, I will talk about our recent advance on this topic. I will cover a new type of between-set distance based on Sparse Approximated Nearest Points (SANP), its formulation as well as optimization and show some experiment results. This work has been accepted as an oral paper in the upcoming CVPR 2011.

Back to top


Robust and Efficient Encoding for Multispectral Palmprint Recognition

  • Zohaib Khan - PhD
  • 11am Friday 13th May
Abstract

The human palm contains rich information which can be used to recognize individuals. Generally, a palmprint contains unique superficial (line) and subsurface (vein) pattern. Unlike conventional imaging which only captures the superficial features, multispectral imaging simultaneously captures both the superficial and subsurface features in a palm, increasing information available for discrimination. The idea has consequently developed interest in multispectral palmprint recognition, where palm line and vein features appearing in multiple bands can be used for recognition. Despite the potential of multispectral information for improved palmprint recognition, the computational complexity is a major challenge which can be addressed by an efficient representation and recognition framework. In this seminar, I will describe 'ContourCode'; a robust and efficient encoding for multispectral palmprint recognition. The framework includes, palmprint region of interest extraction, feature representation, encoding and matching.

Back to top


Novel Bioreactor Systems (Zetos/Chondros) for Ex-Vivo Assessment of Mechanical Behaviour of Bone and Cartilage

  • Professor David Jones - Philipps University Baldingerst, Germany
  • 1.30pm Tuesday 19th April
Abstract

Some tissues such as bone and cartilage are mechanically sensitive. The physical and biochemical mechanisms underlying mechano regulation are complex and at the present time not fully understood. Two devices for applying precise mechanical loads at physiological amplitudes and frequencies were developed, Zetos for bone, a piezo based open loop loading device with a culture system and Chondros, a closed loop controlled voice coil (and recently a single crystal relaxor actuator) all with high resolution force and amplitude sensors were constructed to accurately apply forces and frequencies to bone and cartilage ex-vivo constructs. For bone a large body of evidence suggests that many mechano sensing (and hormonal) mechanisms can be adequately investigated using these ex-vivo bioreactor systems. An infra red imaging device was constructed to allow μCT, or better, resolution of bone microarchitecture and cell activity (presently 5μm resolution xy 3μm z) without any harmful side effects. Using these devices allows real time imaging of structural changes of bone microarchitecture and bone cell distributions.

About the Presenter

Director Institute for Experimental Orthopaedics and Biomechanics Philipps University Baldingerst, 30033 Marburg, Germany Hessen Centre of Excellence for Bioengineering Web: http://www.med.uni-marburg.de/eobm/

Back to top


Retaining both Privacy and Utility in Published Network Traces

  • Dr Chris McDonald - CSSE
  • 11am Friday 25th March
Abstract

Research into computer networks, both wired and wireless, advances more quickly and effectively if researchers share not just their ideas, but also their research tools and data. However, with the increased appreciation of the dangers of sharing private information, it is difficult for network researchers to share their network trace data outside of their institutions. This is a similar problem to that encountered in the health industry, where patient data must first undergo both anonymization and sanitization to remove identifying details before being released to independent researchers. The challenge with robust network trace sanitization is that the identification of network objects must be prevented while retaining the research utility of the sanitized results. This results in a tradeoff being made between privacy and utility. In this seminar I will describe work recently undertaken at Dartmouth College to design and implement a software framework to accommodate most of the currently available privacy and utility metrics for network trace sanitization. The framework provides a set of APIs for analyzing privacy/utility tradeoffs by comparing the changes in privacy and utility levels of a trace for various sanitization operations. Case studies and quantitative evaluation on wireless-network traces will be presented.

Back to top


Fast and Robust Human Recognition Using 2D and 3D Ear and Face Images

  • Dr Syed Islam - UWA
  • 11am Friday 18th February
Abstract

Biometric human recognition is rapidly gaining popularity due to breaches of traditional security systems and the lowering cost of sensors. This talk presents a fast, fully automatic and robust approach for recognition of people from their 2D and 3D ear and face images. At first, the ear shape is detected from the 2D profile image using a Cascaded AdaBoost-based detector. The corresponding 3D ear data are then extracted from the co-registered range image. Face data are detected and extracted from the 3D frontal face scan. Local features are automatically extracted from both the modalities for representation and recognition via construction of a rejection classifier, extraction of a minimal region with feature-rich data points and finally, computation of the initial transformation for matching. The Iterative Closest Point algorithm (ICP) is used for fine matching of the short-listed candidates with the minimal datasets. Fusion of L3DF-based and ICP-based matching scores from the two modalities is performed using a weighted sum-rule. A novel approach of fusing these modalities at feature-level is also proposed. The proposed multimodal biometric approaches significantly outperform current unimodal approaches especially for non-neutral facial expressions.

About the Presenter

Dr. Syed Islam obtained his PhD in Computer Engineering from the School of Computer Science and Software Engineering at The University of Western Australia (UWA) on September 16, 2010. His PhD thesis titled “Human Recognition Using Local 3D Ear and Face Features” was awarded the ‘Award of Distinction’ and was also nominated by his School for the Australasian Distinguished Doctoral Dissertation, 2010. He also completed an MSc in Computer Engineering from King Fahd University of Petroleum and Minerals (KFUPM) in 2006 with a thesis on augmenting a telerobotic stereovision system using graphical overlays. Prior to the MSc, he obtained a Bachelor with Honours in Electrical and Electronic Engineering with substantial proportion of computer courses. So far, he has published or submitted for publication 25 scholarly articles in peer-reviewed journals and conferences including ACM Computing Surveys, IEEE Trans. on Pattern Analysis and Machine Intelligence and International Journal of Computer Vision, which are highly ranked in his area of expertise. He worked as a Lecturer in Computer Science and Engineering Department in the Asian University of Bangladesh from July 9, 2001 to Feb 28, 2003. He also performed several research and teaching assistance jobs in KFUPM and UWA. Currently, he is working as a Postdoctoral Research Fellow in the School of Dentistry at UWA in a project for facial asymmetry analysis using 3D computer graphics.

Back to top

 

School of Computer Science and Software Engineering

This Page

Last updated:
Wednesday, 13 February, 2013 8:23 AM

http://web.csse.uwa.edu.au/2095126