Dr Zahid Islam's Home Page (Data Mining)

Dr Md Zahidul Islam is a Lecturer in Computing, at the School of Computing and Mathematics, Faculty of Business, Charles Sturt University. His main research interests are in Data Mining, Classification and Clustering algorithms, Missing value analysis, Outliers detection, Data Cleaning and Preprocessing, Privacy Preserving Data Mining, Intelligent Decision Support System, Information Flow Control for Information Security, Security of Sensor Networks, Impact of IT on Society (Digital Divide), Graph Theory, Hydroinformatics and so on.

Zahid has received his Bachelors degree in Engineering from Rajshahi University of Engineering and Technology, Bangladesh, Graduate Diploma in Information Science from The University of New South Wales, Australia and PhD in Computer Science (Privacy Preserving Data Mining) from the University of Newcastle, Australia. During his PhD candidature, he was awarded “Faculty of Engineering and Built Environment Postgraduate Research Prize in the Discipline of Computer Science and Software Engineering” in 2005. He has published several peer reviewed publications in various research areas such as Data Mining, Privacy Preservation in Data Mining, Graph Theory, Ethics and Privacy issues of Social Network Site Users Due to Data Mining.

Zahid is a member of Centre for Research in Complex Systems (CRiCS). Along with various colleagues, he has received Faculty of Business Research Fellowship in 2013 worth $40,000, COMPACT Funding 2012 of $9124 for research on clustering, Research Center Fellowship in 2012 worth $40,000, Research Center Fellowship in 2009 worth $40,000, Faculty COMPACT Fund of $18,249 for research on Data Cleansing, COMPACT Fund of $7,000 for research on Data Mining Threats on SNS users,CRiCS Seed grant of $10,000 for his research on intelligent business decision support system, CSU Small Grant of $6,000 for research on classification algorithm, and Faculty Seed Grant of $3,000 for research on a novel clustering technique. His main research interests are in Data Mining algorithms, Fuzzy data mining, Privacy Preservation in Data Mining, Decision Support Systems, Hydroinformatics, Information Flow Control, Security of Sensor Networks, Graph Labelling, Ethics, and Solution to Privacy Threats by Data Minig on Social Network Site Users.

Zahid is currently supervising and cosupervising a number of PhD students who are working in data mining, clustering, classification, data cleansing, privacy issues in data mining, decision support system, Hydroinformatics, Information Flow Control and Managing Information Technology Network and Culture Transition.
A few more PhD students are starting from the next session.
Any potential PhD students are encouraged to contact him via email at zislam@csu.edu.au too see possibilities of scholarships and discuss research interests.

He has teaching experiences at the University of Newcastle and Charles Sturt University, Australia.

My Teaching Philosphy

I am exploring various teaching pedagogy to find the one that suits the best for me. I change my teaching approach from subject to subject. Sometimes, I concentrate to give students as much information as possinble. However, in some other situations I prefer to allow my students to explore, share and learn by themselves.

For example, in a subject like data mining, graph theory and networking students need to know the theory in the background. In this type of subjects I take a research type approach for my students where my role as a lecturer is to provide them the initial knowledge about a topic. I then give them an opportunity to figure out the advantages and disadvantages of the topic, and a possible solution of the problem. The possible solution leads them towards the next topic.

After the initial introduction my role is to guide them towards the right direction. In a lecture room I allow them to work in groups for sometime so that every group can come up with a possible solution.

For example, I provide the initial information and a general introduction if I am teaching encryption techniques in Cryptography or Network Security subject. I then encourage my students to figure out the advantages and disadvantages of encryption techniques such as Playfair Cipher and Transposition Cipher. Once they figure out the drawbacks of these techniques I engage them to work out a possible solution. Once they reach a level then I provide the next level of information so that they can proceed further. We continue this in a very friendly class room environment until they reach the state of the art. I found this teaching technique as very effective to make students interested and engaged in a topic. I think this way they also learn something that they remember for long time. They get a feeling that as if they invented the state of the art technique.

However, in a practical type subject (such as web development and programing language) I encourage them to learn and code by themselves to enjoy their work. I prefer to show them example of what can be done and ask them to do similar things.

Top of page

Teaching Responsibilities

Session 1, 2012

• ITC106 Programming Principles Bathurst Internal
• ITC106 Programming Principles Bathurst Distance
• ITC303 Software Engineering Project 1 Bathurst Internal
• ITC303 Software Engineering Project 1 Bathurst Distance

Semester 2, 2011

• ITC242 Introduction to Data Communications
• ITC431 Computer Networks

Semester 1, 2011

• ITC161 Introduction to Information Technology
• ITC331 Ethics and Professional Practice
• ITC514 Network and Security Administration using Linux
• ITC555 Linux Networking and Security

Semester 2, 2010

• ITC114 Database Management Systems
• ITC242 Introduction to Data Communications
• ITC431 Computer Networks

Semester 1, 2010

• ITC161 Introduction to Information Technology
• ITC331 Ethics and Professional Practice
• ITC514 Network and Security Administration using Linux
• ITC555 Linux Networking and Security

Semester 1, 2009

• ITC570 Special Subject in IT - Data Mining
• ITC105 Communication and Information Management
• ITC532 IT Specialisation Project
• ITC233 Network Engineering 1
• ITC493 IT Project Management

Trimester 2, 2009

• ITC514 Network and Security Administration using Linux

Semester 2, 2009

• ITC 242 Introduction to Data Communications
• ITC 431 Computer Networks
• ITC 499 IT Project Proposal

Other Teaching Experiences

• ITC 518 Principles of Programming using C#
• SENG 4420 Software Architecture
• COMP 1050/6050 Internet Communications
• SENG 3100 Advanced Software Process
• Administrative Responsibilities
• Associate Course coordinator BIT
• Associate Course coordinator BIT/BBus

Research

My Research Interest

• Data Mining,
• Classification Algorithm,
• Clustering Algorithm,
• Missing Value Imputation,
• Outliers Detection,
• Data Cleansing and Preprocessing,
• Decision Forest Algorithm,
• Decision Tree Algorithm,
• Intelligent Decision Support System,
• Application of Data Mining such as Water Demand Prediction and Property Valuation,
• Privacy Preserving Data Mining,
• Privacy Preserving Multi-party Secure Data Mining,
• Data Mining threats on Privacy of Online Community,
• Trust Model,
• Security of Sensor Networks,
• Impact of IT on Society (such as Digital Divide),
• Hydroinformatics,
• Graph Labelling,
• and so on.

Top of page

My Research Focus

Zahid’s PhD topic was “Privacy Preservation in Data Mining through Noise Addition”. His main research interests are in privacy in data mining, various data mining techniques such as Classification and Clustering algorithms, and application of data mining techniques such as water demand forecasting in an irrigation area.

Along with various colleagues, he has received Faculty of Business Research Fellowship in 2013 worth $40,000, COMPACT Funding 2012 of $9124 for research on clustering, Research Center Fellowship in 2012 worth $40,000, Research Center Fellowship in 2009 worth $40,000, Faculty COMPACT Fund of $18,249 for research on Data Cleansing, COMPACT Fund of $7,000 for research on Data Mining Threats on SNS users,CRiCS Seed grant of $10,000 for his research on intelligent business decision support system, CSU Small Grant of $6,000 for research on classification algorithm, and Faculty Seed Grant of $3,000 for research on a novel clustering technique. His main research interests are in Data Mining algorithms, Fuzzy data mining, Privacy Preservation in Data Mining, Decision Support Systems, Hydroinformatics, Information Flow Control, Security of Sensor Networks, Graph Labelling, Digital Divide etc.

Zahid is currently supervising and cosupervising several PhD students who are working in Data Cleansing, Data Pre-processing, Clustering algorithms, Application of Data Mining, Hydro-informatics, Intelligent Decision Support Systems. Zahid is always interested in any new research topics or areas in addition to his main areas of interest.

Potential PhD students are welcome to contact him via email at zislam@csu.edu.au to discuss possibilities of any scholarships.

Top of page

Currently Principal Supervisor of the following PhD/DIT/Honours Students

1. Md Anisur Rahman - PhD.
   
2. Md Geaur Rahman - PhD.
   
3. Samuel Fletcher - PhD.
   
4. Md Nasim Adnan - PhD.
   
5. Abul Hashem Beg - PhD.
   
6. Peter Hough - Honours
   

Grants and Awards:

• Seed Funding from Care West, Australia with Prof Mark Morrisson, Dr Oliver Burmeister, Dr Maree Bernoth, Dr Debra Da Silva, and Dr Bhanugopan Ramudu - $25,000
• Faculty of Business Research Fellowship 2013 - $40,000
• COMPACT Funding for a novel clustering technique, with Prof Bossomaier, Prof Estivill-Castro, and A/Prof Brankovic - $9124.
• Research Center Fellowship 2012 from Center for Research in Complex Systems (CRiCS) - $40,000
• Charles Sturt University Research Infrastructure Block Grants (RIBG), 2012 for "Abnormal event detection using eye tracker technology," with Dr Manoranjan Paul, Prof Junbin Gao, Dr Michael Antolovich, and Prof Terry Bossomaier - $43,000.
• COMPACT Fund 2011 for a research on Data Cleansing and Data Pre-processing, with Prof Bossomaier and Prof Gao - $18,249.
• COMPACT Fund 2011 for research on Data Mining threats on Privacy of Social Network Site (SNS) users, with Dr Al-Saggaf - $7,000.
• CRiCS Special Grant for an Intelligent Decision Support System research - $10,000.
• CRiCS Grant for ARC Linkage Grant Preparation - $4,000.
• Research Center Fellowship at IC Water 2009 - $40,000.
• Faculty Seed Grant from the Faculty of Business, CSU for conducting a research on a novel clustering technique, 2009 - $3,000.
• CSU Small Grant for conducting a research on a novel decision tree classification algorithm, 2009 - $6,000.
• Faculty of Engineering and Built Environment Postgraduate Research Prize in the Discipline of Computer Science and Software Engineering, University of Newcastle, Australia,2005.

Reviewer and Committee Member:

• Reviewer of the Journal of Research and Practice in Information Technology 2008.
• Reviewer of Knowledge-Based Systems, 2009
• Reviewer of the 1 st International Workshop on Quality of Security (QoSec) in Wireless Sensor Networks http://scm.csu.edu.au/qosecwsn09, to be held in conjunction with the 3 rd International Conference on Network & System Security (NSS 2009) http://nss.cqu.edu.au, October 19-21, 2009, Gold Coast, Australia.
• Reviewer of Knowledge-Based Systems, 2010
• Reviewer and PC Member of Eighth ACS/IEEE International Conference on Computer Systems and Applications (AICCSA’ 2010) http://www2.lifl.fr/AICCSA2010/ to be held in Hammamet, Tunisia in May 2010.
• Reviewer of International Conference on Advances in Electrical Engineering (ICAEE), 2011
• Reviewer of Journal of King Saud University - Computer and Information Sciences 2012.
• Reviewer of Knowledge-Based Systems, 2011
• Reviewer of Knowledge-Based Systems, 2012
• Reviewer of Knowledge-Based Systems, 2013
• Member of International Advisory Committee for IEEE-sponsored International Conference on Advances in Electrical Engineering (ICAEE), 2011
• Member of International Advisory Committee of International Conference on Advances in Electrical Engineering (ICAEE), 2011
• Member International Program Committee of the 10th International Conference on E-business (iNCEB2011), will be hold at Asia Hotel, Bangkok, Thailand.
• PC Member of International Conference on Computer Science and Information Technology, Indonesia. (CSIT-2013). http://amcs.co/csit2013/
• PC Member of Australasian Data Mining Conference, (AusDM 2013), Canberra, Australia. http://ausdm13.togaware.com/
• Member of the International Program Committee (IPC) of the 2013 International Conference on Advances in Electrical Engineering, (ICAEE 2013) http://www.icaee.net/index.php
• PC Member of 3rd International Workshop on Applications and Technologies in Information Security (ATIS 2013), Sydney, Australia. http://www.deakin.edu.au/sebe/it/cyberspace-security/

Selected Publications

Please feel free to contact me if you are interested to have a look at any of these published papers.

PhD Thesis:

• Islam, M. Z. (2008): Privacy Preservation in Data Mining through Noise Addition, PhD thesis in Computer Science, School of Electrical Engineering and Computer Science, The University of Newcastle, Australia. PhD08

Book Chapter:

• Brankovic, L, Islam, M. Z. and Giggins H (2007): Privacy-Preserving Data Mining, Security, Privacy and Trust in Modern Data Management, Springer, Editors Milan Petkovic and Willem Jonker ISBN: 978-3-540-69860-9, Chapter 11, pg. 151-166.

Refereed Journal Articles:

• Khan, M. A., Islam, M. Z., and Hafeez, M. (2013): Irrigation Water Requirement Prediction through Various Data Mining Techniques Applied on a Carefully Pre-processed Dataset, Journal of Research and Practice in Information Technology, (Accepted) (ERA Rank B journal)
• Rahman,M. G., and Islam, M. Z. (2013): FIMUS: A Framework for Imputing Missing Values Using Co-appearance, Correlation and Similarity Analysis, Knowledge-Based Systems,(Under Review)(Web of Science Rank Q1)
• Rahman,M. A., and Islam, M. Z. (2013): A Novel Clustering Approach: Density Based Initial Seeds Following Attribute Weights that are Automatically Determined or Manually Assigned by Users According to Their Purpose of Clustering, Knowledge-Based Systems,(Under Review)(Web of Science Rank Q1)
• Rahman,M. G., and Islam, M. Z. (2013): Missing Value Imputation Using Decision Trees and Decision Forests by Splitting and Merging Records, Knowledge-Based Systems,(Under Review)(Web of Science Rank Q1)
• Al-Saggaf, Y., and Islam, M. Z. (2012): Data Mining and Privacy of Social Network Sites' Users: An Empirical Study , First Monday, (under review) (Scopus SJR Rank Q1)
• Rahman, M. G., and Islam, M. Z. (2012): A Framework for Missing Value Imputation using a Novel Fuzzy Expectation Maximisation Algorithm and a Fuzzy Clustering Approach , IEEE Transactions on Knowledge and Data Engineering.(under review)(ERA Rank A)
• Fletcher, S., and Islam, M. Z. (2012): Noise Addition and Data Quality Evaluations using Decision Forests for Privacy Preserving Data Mining , Journal of Research and Practice in Information Technology, (under review) (ERA Rank B)
• Al-Saggaf,Y., and Islam, M. Z. (2012): Privacy in Social Network Sites (SNS) - the threats from Data Mining, Ethical Space: The International Journal of Communication Ethics, Vol. 9, Issue 4, pg. 32 - 40, ISSN 1742-0105. (available at http://journals.communicationethics.net/index.php )(ERA Rank A)
• Islam, M. Z. (2012): EXPLORE: A Novel Decision Tree Classification Algorithm , Data Security and Security Data, Springer, Berlin/Heidelberg (2012), LNCS Vol. 6121, L.M. MacKinnon (Ed.) ISBN 978-3-642-25703-2, pg. 55-71.
• Islam, M.Z., and Brankovic, L.(2011): Privacy Preserving Data Mining: A Noise Addition Framework Using a Novel Clustering Technique, Knowledge-Based Systems Vol. 24, Issue 8, ISBN 0950-7051,(December 2011) pg. 1214-1223, (DOI: 10.1016/j.knosys.2011.05.011) (Impact Factor: 2.42)(Web of Science Rank Q1)
• Zia, T.A., Al-Saggaf, Y., Islam, M. Z., Zheng, L., and Weckert, J. (2009): The Digital Divide in Asia: Cases from Yemen, Bangladesh, Pakistan and China, Journal of Information Ethics (JIE). Vol. 18, No. 2 (Fall 2009). McFarland & Company, Inc. Publishers, pg. 50 - 76. (ERA Rank B)

Refereed Conference Proceedings:

• Rahman, M. G., and Islam, M. Z. (2013): Data Quality Improvement by Imputation of Missing Values, In Proc. of the 2013 International Conference on Computer Science and Information Technology, Yogyakarta, Indonesia, 16 - 18 June, 2013. (Accepted)
• Rahman, M. A., and Islam, M. Z. (2012): CRUDAW: A Novel Fuzzy Technique for Clustering Records Following User Defined Attribute Weights , In Proc. of the 10th Australasian Data Mining Conference (AusDM 12), Sydney, Australia. December 4 - 7, 2012, CRPIT, Vol. 134, Zhao, Y., Li, J., Kennedy, P., and Christen, P. Eds., ACS, pg. 27 - 42. (ERA Rank B)
• Khan, M. A., Islam, M. Z., and Hafeez, M. (2012): Evaluating the Performance of Several Data Mining Methods for Predicting Irrigation Water Requirement , In Proc. of the 10th Australasian Data Mining Conference (AusDM 12), Sydney, Australia. December 4 - 7, 2012, CRPIT, Vol. 134, Zhao, Y., Li, J., Kennedy, P., and Christen, P. Eds., ACS, pg. 199 - 208. (ERA Rank B)
• Bossomaier, T., Islam, M. Z., Duncan, R., and Gao, J. (2012): Simulation of House Prices for Improved Land Valuation , In Proc. of the 24th European Modeling & Simulation Symposium (EMSS 12), Vienna, Austria. September 19 - 21, 2012.
• Rahman, M. G., Islam, M. Z., Bossomaier, T., and Gao, J. (2012): CAIRAD: A Novel Technique for Incorrect Records and Attribute-values Detection , In Proc. of IEEE International Joint Conference on Neural Networks (IJCNN 12), Brisbane, Australia. June 10 - June 15, 2012, pg. 2190-2199.(ERA Rank A)
• Islam, M. Z. and Giggins, H. (2011): Knowledge Discovery through SysFor: A Systematically Developed Forest of Multiple Decision Trees , In Proc. of the Ninth Australasian Data Mining Conference (AusDM 11), Ballarat, Australia. December 01 - December 02, 2011. CRPIT, 121. Vamplew, P., Stranieri, A., Ong, K.-L., Christen, P. and Kennedy, P. J. Eds., ACS. 195-204 (ERA Rank B)(available at http://crpit.com/vol121.html)
• Rahman, M. G. and Islam, M. Z. (2011): A Decision Tree-based Missing Value Imputation Technique for Data Pre-processing , In Proc. of the Ninth Australasian Data Mining Conference (AusDM 11), Ballarat, Australia. December 01 - December 02, 2011, CRPIT, 121. Vamplew, P., Stranieri, A., Ong, K.-L., Christen, P. and Kennedy, P. J. Eds., ACS. 41-50 (ERA Rank B)(available at http://crpit.com/vol121.html)
• Rahman, M. A. and Islam, M. Z. (2011): Seed-Detective: A Novel Clustering Technique Using High Quality Seed for K-Means on Categorical and Numerical Attributes , In Proc. of the Ninth Australasian Data Mining Conference (AusDM 11), Ballarat, Australia. December 01 - December 02, 2011, CRPIT, 121. Vamplew, P., Stranieri, A., Ong, K.-L., Christen, P. and Kennedy, P. J. Eds., ACS. 211-220 (ERA Rank B)(available at http://crpit.com/vol121.html)
• Khan, M. A., Islam, M. Z., and Hafeez, M. (2011): Irrigation Water Demand Forecasting - A Data Pre-processing and Data Mining Approach Based on Spatiotemporal Data , In Proc. of the Ninth Australasian Data Mining Conference (AusDM 11), Ballarat, Australia. December 01 - December 02, 2011, CRPIT, 121. Vamplew, P., Stranieri, A., Ong, K.-L., Christen, P. and Kennedy, P. J. Eds., ACS. 183-194 (ERA Rank B) (available at http://crpit.com/vol121.html)(Best Student Paper Award)
• Islam, M. Z. (2010): EXPLORE: A Novel Decision Tree Classification Algorithm , In Proc. of the 27th British National Conference on Databases (BNCOD 2010), June 29- July 01, 2010, Dundee, Scotland. L.M. MacKinnon (Ed.) 55-71. (ERA Rank B)
• Zia, T.A., and Islam, M. Z. (2010): Communal Reputation and Individual Trust (CRIT) in Wireless Sensor Networks , In Proc. of the 5th International Conference on Availability, Reliability and Security (ARES 2010), Published by IEEE Computer Society, February 15-18, 2010, Krakow, Poland, 347-352. (ERA Rank B)
• Islam, M. Z. and Brankovic, L (2005): DETECTIVE: A Decision Tree Based Categorical Value Clustering and Perturbation Technique in Privacy Preserving Data Mining, In Proc. of the 3rd International IEEE Conference on Industrial Informatics (INDIN 2005), 10-12 August, Perth, Australia.
• Alfalayleh, M., Brankovic, L., Giggins, H and Islam, M. Z. (2004): Towards the Graceful Tree Conjecture: A Survey , In Proc. of AWOCA 2004, 7-9 July, Ballina, Australia. (ERA Rank B)
• Islam, M. Z. and Brankovic, L. (2004): A Framework for Privacy Preserving Classification in Data Mining,In Proc. of Australasian Workshop on Data Mining and Web Intelligence (DMWI 2004), Dunedin, New Zealand, CRPIT, 32, Purvis, M., Ed. ACS, 163-168.
• Islam, M. Z., Barnaghi, P. M. and Brankovic, L.(2003): Measuring Data Quality: Predictive Accuracy vs.Similarity of Decision Trees, In Proceedings of the 6 th International Conferenceon Computer & Information Technology (ICCIT 2003), Dhaka, Bangladesh, Vol. 2, 457-462. (ERA Rank C)
• Islam, M. Z. and Brankovic, L. (2003): Noise Addition for Protecting Privacy in Data Mining, In Proceedings of the 6 th Engineering Mathematics and Applications Conference (EMAC 2003), Sydney, Australia, 85-90 .

Current Research Projects

Automatic and Natural Clustering of Records ($9124 Faculty COMPACT Fund 2012): Dr Islam, Prof Bossomaier, Prof Estivill-Castro, A/Prof Brankovic

• 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.

Data Cleansing and Data Pre-processing Techniques($18,249 Faculty COMPACT Fund 2011): Dr Islam, Prof Bossomaier, Prof Gao

• 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.
  
  

Data Mining threats on Privacy of Social Network Site (SNS) users ($7,000 Faculty COMPACT Fund 2011): Dr Al-Saggaf, Dr Islam

• 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.
  
  

Novel Data Mining Techniques for an Intelligent Business Decision Support System ($10,000 CRiCS Seed Grant 2010): Dr Islam

• Due to the recent development of information processing technology and storage capacity businesses typically collect huge amount of data nowadays. Data are not useful unless necessary information is extracted from them. Various data mining tasks including data cleaning, classification, clustering, and prediction are usually applied on collected data for knowledge discovery.
  
  In this project we are developing an intelligent Decision Support System (DSS) that will integrate available data of an organisation by automatically extracting information from textual data through text data mining, detecting and correcting any corrupt (incorrect) data, and imputing all missing values. Our intelligent decision support system will be capable of grouping similar records, extracting multiple sets of patterns (instead of a single set of patterns or logic rules) through classification, and predicting future with very high accuracy. While grouping similar records our DSS will give a user huge flexibility to assign different weights on different attributes and thereby experience different groups of similar records in order to explore various patterns. The DSS will also provide a classification algorithm for extracting interesting patterns that are generally ignored by existing algorithms. For faster analysis of huge amount of real time data, our DSS will use GPGPU for parallel processing of our algorithms. GPU (instead of CPU) can process independent calculations in parallel using a single kernel on many records requiring similar calculations. We will use CUDA architecture for parallel computation.
  
  As a result the DSS can be used in various purposes such as diagnosis and prevention of diseases, behaviour analysis of equipments and predict any future breakdown, making decision on bank loan applications, and identification of suspicious tax returns - just to name a few.
  
  

A Novel Decision Tree Classification Algorithm ($6,000 CSU Small Grant 2009):Dr Islam

• 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.
  
  These patterns can then be used in our original noise addition techniques and similarity evaluation of categorical values. My current Seed Grant project (2009) is for developing clustering technique where as the aim of this Small Grant project is to develop a classification technique. Although these two studies are significantly different to each other, they will both be used in our noise addition framework.
  
  

A Novel Clustering Technique for Categorical and Numerical Values ($3,000 Faculty of Business Seed Grant 2009): Dr Islam

• Advances in information processing technology and storage capacity have enabled collection of huge amount of data for various data analyses. Data mining techniques such as classification are often applied on these data to extract hidden information. During the whole process of data mining these data get exposed to several parties which can potentially lead to breaches of individual privacy.
  
  During our previous research, we have presented a framework that uses a few novel noise addition techniques for protecting individual privacy while maintaining a high data quality. We added noise to all attributes, both numerical and categorical. Noise addition techniques developed for numerical attributes are not suitable for noise addition to categorical attributes, due to the absence of natural ordering in categorical attributes. Therefore, we presented a novel technique for clustering categorical values, and used it for noise addition to those values. We also proposed an extension of the technique for clustering records (not just attribute values) having categorical and/or numerical attributes. Our initial experiments indicate suitability of these techniques in clustering categorical values and in noise addition to these values. However, the clustering techniques (specially the one for clustering records) need to be sufficiently experimented and improved. They need to be compared with existing techniques in order to demonstrate their advantages.
  
  The aim of this project is to carry out extensive experiments to evaluate the performance of the proposed clustering techniques by comparing them with existing techniques such as CACTUS, ROCK and QROCK. The study also aims to improve the proposed clustering techniques. Their usefulness for being used in the noise addition framework will also be evaluated through the data quality and level of privacy in a perturbed data set.
  
  

A Privacy Preserving Data Mining Technique (Faculty of Business Research Assistants Support for Honours Students):Dr Islam

• Due to the development of data collection and storage facilities, organisations these days collect a huge amount of data in almost every sector of life. Collected data (such as a patient dataset of a hospital) typically contain sensitive individual information. While the data can be useful in research, business analysis, decision making and prediction an organisation (such as a hospital) often is restricted to share and disclose the collected data due to potential breach of individual privacy. This restriction limits the usefulness of sophisticated data mining techniques in exploring patterns and other hidden information (such as causes of a disease and possible prevention techniques) from the collected data. We aim to develop techniques to ensure the privacy of data subjects while releasing the data for sharing among interested parties for knowledge discovery.
  
  We previously presented a complete framework for noise addition to all attributes of a data set in order to protect individual privacy in a data set while maintaining its original data quality. That is, we added noise to both numerical and categorical attributes in such a way so that the original patterns are preserved in a perturbed data set. We also presented an extended framework, which is also capable of incorporating previously proposed noise addition techniques that maintain the statistical parameters including correlations among attributes of a data set. Thus the perturbed data set can be used not only for classification but also for other statistical analysis.
  
  We tested the framework and extended framework on different data sets, and compared our frameworks with some other noise addition techniques, through data quality of perturbed data sets. Data quality of a perturbed data set was evaluated through a few parameters such as the similarities of decision trees obtained from the original data set and the perturbed data set, prediction accuracy of a decision tree obtained from the perturbed data set, and the correlation matrices produced from the perturbed data set and the original data set. We also presented a technique for security analysis of a data set. An initial experiment indicates the existence of higher level of security in a data set perturbed by our framework than the security level in the original data set.
  
  In this project we are working on further improvement of our noise addition techniques and security analysis.
  
  

Privacy Preserving Multi-party Data Mining

• Often multiple parties (such as the Commonwealth Bank and the Westpac Bank) intend to perform data mining on their combined dataset in order to extract more useful knowledge than the knowledge that can be extracted through data mining on their individual datasets separately. However, they typically hesitate to release their datasets to other parties due to the obligations to their customers for preserving individual privacy. Moreover, a party may not also want to disclose sensitive rules (such as they never approve loans to a particular group of people like the residents of a suburb simply due to the fact that they live there) to other parties. Such a disclosure can easily damage their reputation and image publicly. In this project we aim to develop techniques that will help the parties to effectively mine their combined data while preserving individual privacy and sensitive rules.

SysFor: A Systematically Developed Forest of Multiple Trees: Dr Islam

• Decision tree based classification algorithms like C4.5 and Explore build a single tree from a data set. The two main purposes of building a decision tree are to extract various patterns/logic-rules existing in a data set, and to predict the class attribute value of an unlabeled record. Often a set of decision trees, rather than just a single tree, are also generated from a data set. The set of multiple trees, when used wisely, typically have better prediction accuracy on unlabeled records. In this project we present a novel technique for building a set of multiple trees called SysFor. Our initial experimental results demonstrate that SysFor is suitable for multiple pattern extraction and knowledge discovery. Moreover, it also has higher prediction accuracy than a couple of existing champion techniques.

I am extremely interested in group research. Please feel free to contact me if you are interested in any of the above projects and/or have ideas on new projects. I have also worked with my wonderful colleagues in reserch areas such as Digital Divide, Security of Sensor Networks, Hydroinformatics and Graph Theory - just to show a diversity of my research interest, especially when it is in a group.

Potential PhD students are also encouraged to contact me at my email address (zislam@csu.edu.au) to discuss research topics and possible scholarships.

Latest News

Visits of our research collaborators.

Our research group regularly invites collaborators and researchers from all over the places. In 2012 Prof Vladimir Estivill-Castro of Griffith University and A/Prof Ljiljana Brankovic of Newcastle University visited us.

A/Prof Ljiljana Brankovic visited us in Bathurst, 2012. A wonderful moment with my PhD supervisor and a few PhD students.

I am introducing A/Prof Brankovic in a seminar during her visit

Attendeding Various Conferences on Data Mining

Members of our group regularly attend high quality international conferences. In 2012 we attended AusDM 2012, IJCNN 2012, and ACSW 2012.

A wonderful moment with my PhD students in AusDM 2012 at Sydney
On behalf of Mahmood Khan I am receiving the best student paper award for our paper with Mahmood at AusDM 2011 at Ballarat	My talk on our novel Decision Forest technique at AusDM 2011 at Ballarat
With my PhD students at AusDM 2011	A happy research team. A very nice moment at AusDM 2011 at Ballarat