Charles Sturt University
Charles Sturt University

Computational Intelligence and Robotics

Research Area leader:

Dr Michael Antolovich

Dr Michael Antolovich

  • Overview
  • Members
  • Projects
  • Current Students
  • PhD/DBA Topics
  • Honours

Overview

The CIR Research Area (RA) is one of the long term successful and focused research areas within the Faculty of Business, Justice and Behavioural Sciences. The researchers working in this research area are key members of the Computational Intelligence Research Group within the School of Computing and Mathematics, and their research is well aligned with the major themes of CRiCS (Centre for Research in Complex Systems).

The aim of this RA is to address real world problems in relation to smart information technologies in industries and Australian public sectors with the mission to serve communities through advancing frontier technology. This RA is well aligned with the national research priority Frontier Technologies for Building and Transforming Australian Industries. Under this RA, the group members are to devote their efforts to the four major themes with a wide spectrum of research concerning smart information use covering:

  • Intelligent Computation (IC)
  • Image, Video and Vision Research (IVVR)
  • Robotics and Automatics (RA)
  • Data Analytics and Making Sense of Data (DAMSD)

These research themes develop a set of innovative methodologies and techniques for smart information processing and system building for a broad range of practical applications for industries, including mining, engineering, management and health, and public sectors.

The research group accommodates a range of members from earlier career researchers, established researchers and some internationally recognised researchers. Several members have been actively engaged in conducting CRC Projects, ARC Discovery and Linkage Projects, Industry Funded Projects, and Projects funded by international funding bodies such as National Science Foundation of China (NSFC). Most of their research works have been widely published in internationally top ranked journals such as IEEE Transactions on Pattern Analysis and Machine Intelligence, IEEE Transactions on Neural Networks, Neural Computation, Machine Learning, IEEE Transactions on Image Processing, IEEE Transactions on Circuits and Systems for Video Technology, IEEE on Multimedia and major international conference proceedings such as ICIP, ICASSP, ICME, CEC, GECCO and IJCNN.

The group enjoys several dedicated lab facilities established by CRiCS. The labs include a 3G Visualization Lab, a Computer Vision Lab, a Newcrest Mining lab, Robotics lab and High Performance Computing Lab. These labs not only enable the group members to conduct their research activities but also help PhD students to do their research.

Members

NameTitle/PositionEmployer if not Faculty of Business, Justice and Behavioural Sciences, CSULocation
Terry Bossomaier Professor   Bathurst
Junbin GaoAdjunct Professor Bathurst
Michael Antolovich Senior Lecturer   Bathurst
Xiaodi Huang Senior Lecturer   Albury
David Tien Senior Lecturer   Bathurst
Maumita Bhattacharya Lecturer   Albury
Michael Kemp Lecturer   Orange
Zahidul Islam Lecturer   Bathurst
Manoranjan Paul Lecturer   Bathurst
Jim Tulip Lecturer   Bathurst
Lihong Zheng Lecturer   Wagga Wagga
Chandana Withana Adjunct Associate Professor Study Group, Sydney Sydney
Sudath Heiyantuduwage Adjunct Lecturer Study Group, Sydney Sydney
Daming Shi Adjunct Professor Harbin Institute of Technology and Middlesex University London, UK Harbin, China
Xia Hong Adjunct Reader University of Reading Reading, UK
Paul Kwan Adjunct Senior Lecturer University of New England Armidale
Zhouchen Lin Adjunct Professor Peking University Beijing, China

Projects

Future Projects

Project Name

Brief  Description

Investigators 

Privacy of Social Network Site (SNS) users: A Data Mining  Solution

The aim of the study is to explore the potential of data mining as a technique that could be used by malicious data miners to threaten the privacy of Social Network  Sites (SNS) users AND provide a technical solution to the problem.

Dr Islam and Dr  Yeslam Al-Saggaf

Learning on Manifolds

The long term project aims at investigating learning algorithms on manifolds in computer vision.

Prof Junbin Gao 

Human Gesture and Behaviours  Recognition

Human uses gestures to depict sign language to deaf people, convey messages in noisy environments, and interface with computer games. The emphasis of the project is on automatic learning of vocabularies of gestures performed by a single user or several different users under simple or complicated environment.  The outcomes of this project can be applied in many interaction and computer vision applications, including computer interaction, video surveillance, rehabilitation and health care.

Dr Lihong Zheng

 

Current Projects

Project Name  Brief  Description Funding Body  Investigators 
Online Learning for Large Scale Structured Data in Complex Situations  This project is to develop Online Learning  algorithms to unlock the potential of such overwhelming data, which can  lift existing applications to a new level. Reducing 1% of the  total fraud loss of $3.5 trillion by research, is 35 billion dollars  return. The advances in fraud detection also make Australia a  safer nation which attracts more overseas investments.  Advancing social networks by research will secure Australia's future  share of the social networks market that has massive potential.  ARC Discovery 

Dr Qinfeng Shi (Uni of Adelaide), Prof  Junbin Gao and Prof Sheng Chen (University of Southampton, UK)

A probabilistic  framework for nonlinear dimensionality reduction algorithms The Twin Measures Framework is a novel platform  for analysing existing dimensionality reduction methods and the  invention of new ones. This research will radically improve  image analysis, with beneficial applications from pharmaceutical drug  design through to border protection. ARC Discovery 

Prof Junbin Gao and Prof Xia Hong (University of Reading, UK)

Multi-view video coding using cuboid data compression   This project investigates novel approaches to  multiview video coding that use new data compression techniques and  explicit occlusion handling.  These new approaches complement  the state-of-the-art, improving interactivity with instantaneous view  change and VCR functionality, reducing encoding complexity and  increasing compression efficiency. ARC Discovery  Prof Manzur Murshed (Monash) and Dr. Manoranjan Paul
Drawbell Boulder  Detection This project extends the  current research on draw point boulder detection by the Mining Research  Laboratory at CSU.MMT3 Consortium Prof Junbin Gao, Dr. Michael Antolovich and Mr. Allen Benter  
Automatic and Natural  Clustering of Records In this study we aim to  further improve our clustering techniques in order to group records in  more meaningful clusters with automatic cluster number selection,  attribute weights for clustering and so on. Clustering results will also  be validated by various existing evaluation techniques and novel  evaluation techniques. Faculty of Business,  Justice and Behavioural Sciences Compact Fund Dr  Islam, Prof Bossomaier, Prof Estivill-Castro, A/Prof Brankovic 
Exploiting the Structural  Consistency of Multi-modal Data. The  project will study the problem of multi-modal learning, and exploit the  inherent complementary visual properties of multi-modal data.  Faculty of Business, Justice and Behavioural  Sciences Compact Fund

Dr Xiaodi Huang and  Dr Lin Wu

Biomedical  Image Analysis

The research  project is to develop an automatic, rotation and scale invariant  segmentation method to detect contour of horse larynx from endoscopic  static image or video for diagnostic as well as surgery procedure under  different lighting conditions considering speed and  accuracy.

CSU International PhD  Scholarship

Dr Lihong Zheng and Prof Junbin Gao

Efficient  Low-Resolution Image Segmentation The similarity  measure criteria play critical role in accurate segmentation. This  project aims at finding an appropriate similarity measure to extract  objects in an image efficiently. Faculty of  Business, Justice and Behavioural Sciences Compact Fund  Dr Lihong Zheng 
Gesture recognition 

This research project is to recognition  human gesture based on depth images. Firstly, the discriminatory informative representation for a gesture in 3D space containing disparity  gesture information will be identified. Then a 2D-MTM (motion trail  model) and a three dimensional motion trail model (3D-MTM) will be built  up and applied to demonstrate the accuracy and effectiveness of the  proposed 3D-MTM for human gesture recognition.

CSU International  PhD Scholarship Dr Lihong Zheng 

Past Projects

Project Name  Brief  Description Funding Body  Investigators 
Imaging Techniques to  Determine Muck Pile Ore Fragment Size In-Situ   We investigate real-time techniques for detection  of larger size ore fragment in muck pile under the production  environment. Newcrest Mining Australia  Dr Michael Antolovich and Prof Junbin Gao

WAREIGS:  Wavelet-Network-Based Augmented-Reality-Enhanced Image-Guided  Surgery

This research will address issues  with the constraint of surgical openings where surgeons cannot see  beyond the exposed surfaces and these limitations are accentuated by the  even greater restrictions of minimally invasive surgery. Limited visibility through "keyholes" during endoscopic procedures and  through small incisions with ever-diminishing sizes increases the need  for intra-operative image guidance. The project aims to develop a system of augmented reality enhanced image guided surgical  surgery, in which the images from cameras are aligned with patient's  physical position at the time of surgery. Such a "see-through" capability makes surgeries safer and more accurate. two  objectives will be achieved: (1) Theoretical research on image  registration based on wavelet networks. (2) Development of augmented reality enhanced image-guided surgery with see-through capability. ACSRF, Australian Government  Prof Junbin Gao, Dr. Michael Antolovich,
Prof Terry Bossomaier , Dr. Manoranjan Paul, Dr. Paul Kwan (UNE) and Professor Daming Shi (University of Middlesex, UK 
3D Video Coding  The project investigates the video coding  technologies for multi-view video codingFaculty of Business, Justice and Behavioural Sciences Compact Fund  Dr Manoranjan Paul 
Image Texture Segmentation  Based on Wigner Distribution in a Fractional Fourier Domain  To apply the proposed WD-FrFT to image texture  segmentation. Since coarseness and directionality are two essential perceptual cues used by the human visual system for  discriminating different textures, we will adopt two corresponding types  of features, spatial - frequency and orientation, which will be  extracted by L1 difference of Gaussian (DOG) filters and L2 wedge  filters, respectively. Faculty of  Business, Justice and Behavioural Sciences Compact Fund  Prof Junbin Gao, Dr. Michael Antolovich, Dr.  Manoranjan Paul, Dr. Jim Tulip and Dr. David Tien 
Efficient Low-Resolution  Image Segmentation

The similarity measure criteria play a critical role in accurate segmentation. This  project aims at finding an appropriate similarity measure to extract  objects in an image efficiently.

Faculty of Business, Justice and Behavioural Sciences Compact Fund  Dr Lihong Zheng and Prof Junbin Gao 
Computational Intelligence  for Anomaly Detection in Networks: An Investigation  This project investigates the use of computational  intelligence (CI) for anomaly detection in computer networks. Also,  comparative performances of existing signature-based and misuse  detection approaches are investigated. Faculty of  Business, Justice and Behavioural Sciences Compact Fund  Maumita Bhattacharya and Dr Tanveer Zia 
Age Care Workforce Reform -  Building Communities of Practice Around the Prevention of Functional  Decline in the Community

This research  seeks to investigate whether improved training and use of technology by  clinicians (support workers) and training of volunteers improves human  resource management outcomes among employees and volunteer carers who  are involved in reducing the rate of functional decline among seniors.  This research involves the use of experiments and pre-post surveys of  subjects. Various data mining techniques are applied on the survey data  for extracting the patterns and information in order to evaluate the  impact of various interventions.

CareWest, Australia Prof Mark Morrison, Dr Oliver  Burmeister, Dr Zahid Islam, Dr Ramudu Bhanugopan and Dr Maree Bernoth 
Review of Support Worker Integration with Functional Decline The main aim  of this project is to suggest reasons and possible remedies of the  functional decline for the employees and care receivers of Hobart District Nursing. We apply data mining and other data analyses on the collected data. Hobart District Nursing, Australia Prof Mark Morrisson, Dr Maree Bernoth, Dr Oliver Burmeister, and Dr. Zahid Islam 
Co-ranking Images and Tags in  a Heterogeneous Network

This  project is to develop novel algorithms that are able to handle dual-relational data over a combined graph for co-ranking images and  tags simultaneously. In particular, a co-ranking scheme for images and their associated tags in a heterogeneous graph will be designed and tested on real datasets, together with a prototype  system.

Faculty of Business, Justice and Behavioural Sciences Compact Fund 

Dr Xiaodi Huang and Dr Lin Wu

Data Cleansing  and Data Pre-processing Techniques

In this study we develop novel data cleansing techniques for improving data quality. The improved data will  then be used in making better decision for an organisation.

We have also developed an Agent Based Modeling (ABM) which is powered by data mining techniques in order to  simulate the property market and thereby predict and assess property  prices. We are also developing various techniques for more acceptable  property valuation

Faculty of Business,  Justice and Behavioural Sciences;Compact Fund  Dr Islam, Prof Terrry Bossomaier, Prof Junbin Gao 
Data Mining threats on  Privacy of Social Network Site (SNS) users The aim of the study is to explore the potential  of data mining as a technique that could be used by malicious data  miners to threaten the privacy of Social Network Sites (SNS) users.  Faculty of Business, Justice and Behavioural  Sciences Compact Fund Dr Yeslam Al-Saggaf and Dr Islam
Learning based Boundary/Contour Extraction   Automatic Number Plate Recognition (ANPR) is  required due to increasing traffic management. This project will deliver  a technology capable of efficiently extracting trustful boundary feature for car number plates. CSU Small Grant  Dr Lihong Zheng 
Efficient Character  Segmentation in ANPR This project will deliver an accurate and efficient technology to cut characters from  located images of car number plates.Faculty Seed Grant Dr Lihong Zheng 
A Novel Decision  Tree Classification Algorithm The aim of this  project is to develop a novel decision tree algorithm that will extract  useful patterns (that are currently ignored by existing algorithms) from  a data set. We study various existing classification algorithms such as  Decision tree algorithms, Neural networks, Bayesian algorithms and  Genetic algorithms. We propose some modifications to existing algorithms  and test the algorithm by applying it on a number of data sets. We  compare the efficiencies of the proposed algorithm with various existing  algorithms such as See 5, J48, and REPTree. The efficiencies are evaluated based on quality of extracted pattern, simplicity of the  trees, Performance and Significance of logic rules. Our initial  experimental results are very encouraging. CSU Small Grant 2009 Dr Islam

Current Students

Student Award Topic Supervisor
Timothy Dobson Hons Representation learning in games technology.

Dr Jim Tulip
jtulip@csu.edu.au
02 6338 4931

Saliya Wimalartne Hons Investigation on a biologically inspired heuristic and its application to computer security. Maumita Bhattacharya
mbhattacharya@csu.edu.au
02 6051 9619

PhD/DBA Topics

The following opportunities exist in Computational Intelligence and Robotics:

SupervisorTopic
Dr Manoranjan Paul
mpaul@csu.edu.au
02 6338 4260
  • Conversion of 2D video to 3D video
  • 3D video coding for real time communications
  • Panic-Driven event detection for video surveillance
Dr Zahid Islam
zislam@csu.edu.au
02 6933 2415
  • Data mining techniques to reduce security and privacy threats
  • Data cleansing and pre-processing to improve data quality
  • Data analysis to predict the future and make better business decisions
Dr Lihong Zheng
lzheng@csu.edu.au
02 6933 2387
  • Biomedical image analysis
Maumita Bhattacharya
mbhattacharya@csu.edu.au
02 6051 9619
  • Meta-heuristics for very high dimensional optimisation problems
Dr Xiaodi Huang
xhuang@csu.edu.au
02 6051 9652
  • E-recommendation algorithms and systems
  • Algorithms for visualization of large graphs

Honours

SupervisorTopic

Dr Manoranjan Paul
mpaul@csu.edu.au
02 6338 4260
Bathurst
Discipline: Computer Science

  • Fall detection
  • Visual attention modelling

Dr Zahid Islam
zislam@csu.edu.au02 69332415
Bathurst
Discipline: Computer Science

  • Equipment failure prediction through artificial neural networks

Dr Lihong Zheng
lzheng@csu.edu.au
02 6933 2387
Wagga Wagga
Discipline: Computer Science

  • Efficient segmentation on low resolution image

Maumita Bhattacharya
mbhattacharya@csu.edu.au
02 6051 9619
Albury-Wodonga
Discipline: Computer Science

  • Biologically inspired meta-heuristics for computer security
  • Computational Intelligence for business decision making

Dr Xiaodi Huang
xhuang@csu.edu.au
02 6051 9652
Albury-Wodonga
Discipline: Computer Science

  • Visualization in Internet of things
  • Visualization in cloud computing