Invited and Accepted Papers

Keynotes

• DATA = NORMAL + ANOMALOUS + NOISE
  Professor Sanjay Chawla

  Abstract: Our world at the micro, macro and personal level is now highly instrumented. A consequence of this instrumentation is that now it is possible to obtain fine-grained data about almost anything of interest. Once we focus on an application or a domain, it is reasonable to assume that much of the data obtained captures the "normal" behavior of the underlying phenomenon. Historically, "knowledge discovery," if any, has been triggered by the non-normal or anomalous part of the data. In this talk I will present some classic examples of data anomalies and how their discovery has changed our understanding of the world. Then I will present a modern and algorithmic viewpoint of anomaly detection as is currently practiced in the data mining community.

  Sanjay Chawla is the Professor of Pattern and Data Mining at the University of Sydney. He served as the Head of the School of Information Technologies from 2008-2011. His research work has appeared in leading international conferences and journals. Along with his students, he has received four best paper awards in the last six years. In 2012, he was the program co-chair of the 16th Pacific-Asia Conference on Knowledge Discovery and Data Mining (PAKDD). He serves on the editorial board of the journal DMKD and is a member of the ACM.
  

• Non-iidness: Coupled Object and Pattern Analysis
  Professor Longbing Cao

  Most of existing data mining algorithms are based on the IID assumption, which treats objects independently from each other. In the real world, objects are either loosely or tightly coupled with each other. For instance, a moving vehicle on the street interacts with the cars before and after it, and the ones on its left and right hand sides if any. In social networks, people interact with each other at different levels for varied purposes. Such interactions, or coupling relationships, are ubiquitous, and spread at various levels, between objects, between attributes describing an object, between attribute values within an attribute. It is crucial to cater for such relations in object analysis.

  On the other hand, the usual patterns identified by data mining are based on independent objects or items. For instance, often a large number of frequent patterns are mined by the existing algorithms, which are often treated as independent with each other. In fact, due to the object coupling relationships, patterns are associated with each other in structural and/or semantic aspects. Pattern relationship analysis is often ignored.

  In this talk, we will explore the needs, challenges, opportunities of analyzing complex object relations and complex pattern relations. On top of a framework for noniid-based coupled object and pattern analysis, several corresponding techniques will be introduced: coupled object analysis to define and quantify the coupling relationships within and between objects and within and between attributes, combined pattern mining to identify a group of patterns coupled by certain relationships. Coupled behavior analysis will be explored to analyse a group of actors’ behaviors. We will show how such new frameworks outperform the classic iid-based data mining framework in terms of handling complex data, behavior, relation, environment and pattern in clustering, frequent pattern mining, and classification. Several real-life applications will be given, such as the identification of group-based market manipulations in stock markets.

  Dr Longbing Cao is a professor of information technology at the University of Technology Sydney (UTS) Australia. He got PhD in Intelligent Sciences and PhD in Computing Science. He is the Founding Director of Advanced Analytics Institute, one of three university research institutes at UTS. AAI is the largest analytics research group in Australia, with the widest and deepest engagement with major local and overseas government and industry organizations in many sectors and domains. Prof Cao is also the Research Leader of the Data Mining Program at the Australian Capital Markets Cooperative Research Centre. He is a Senior Member of IEEE, SMC Society and Computer Society.

  Prof Cao's primary research interests include data mining and machine learning, and artificial intelligence and intelligent systems. He initiated and is leading the research on behavior informatics, domain driven data mining, agent mining, and open complex intelligent systems. He has published 2 monographs, 17 edited books and proceedings, 11 book chapters, and over 150 journal/conference publications. In recent years, his interest is mainly on non-iid based learning and mining, targeting coupled object analysis and pattern relation analysis.

  Coming from a seniorindustry leadership commitment, his following expertise and experience has been widely recognized in the academia, government, business and industry: enterprise applications of intelligent big data analytics and intelligent systems and decisions in areas such as customer analysis, fraud detection, outlier detection, risk management, and operational analysis in domains including social security, taxation, capital markets, online business, banking, telecommunication, health care, insurance, marketing, and so on in the real world.

Accepted Papers

• Yun Sing Koh and Gill Dobbie
  Indirect Weighted Association Rules Mining for Academic Network Collaboration Recommendations
• Yu-Shiang Hung, Kuei-Ling B. Chen, Chi-Ta Yang and Guang-Feng Deng
  Mining cluster-based patterns for elder self-care behavior
• Dr Andrei Kelarev, Andrew Stranieri, John Yearwood, Herbert Jelinek and Jemal Abawajy
  Improving Classifications for Cardiac Autonomic Neuropathy Using Multi-level Ensemble Classifiers and Feature Selection Based on Random Forest
• Muhammad Marwan Muhammad Fuad
  ABC-SG: A New Artificial Bee Colony Algorithm-Based Distance of Sequential Data Using Sigma Grams
• Jonathan Wells, Kai Ming Ting and Chandrasiri Naiwala
  A non-time series approach to vehicle related time series problems
• Mahmood Khan and Md. Zahidul Islam
  Evaluating the Performance of Several Data Mining Methods for Predicting Irrigation Water Requirement
• Lavneet Singh and Girija Chetty
  A Novel Machine Learning Approach for Detecting the Brain Abnormalities from MRI Structural Images
• Endang Djuana, Yue Xu and Yuefeng Li
  Learning Personalized Tag Ontology from User Tagging Information
• Dang Bach Bui, Fedja Hadzic and Michael Hecker
  Application of Tree-structured Data Mining for Analysis of Process Logs in XML format
• Md Anisur Rahman and Md Zahidul Islam
  CRUDAW: A Novel Fuzzy Technique for Clustering Records Following User Defined Attribute Weights
• Kamal Taha
  GOtoGene: A Method for Determining the Functional Similarity among Gene Products
• Muhammad Usman, Russel Pears and Alvis Fong
  Data Guided Approach to Generate Multi-dimensional Schema for Targeted Knowledge Discovery
• Chao Sun, David Stirling, Christian Ritz and Claude Sammut
  Variance-wise Segmentation for a Temporal-Adaptive SAX
• Dinusha Vatsalan and Peter Christen
  An Iterative Two-Party Protocol for Scalable Privacy-Preserving Record Linkage
• T. L. Oshini Goonetilleke and H. A. Caldera
  Mining Life Insurance Data for Customer Attrition Analysis
• Michael Mayo
  Cartesian Genetic Programming for Trading: A Preliminary Investigation
• Jing Zhang, Derek Yu-Hsn Liu, Kok-Leong Ong, Zhijie Li and Ming Li
  Detecting Topic Labels for Tweets by Matching Features from Pseudo-Relevance Feedback
• Hamid Ghous, Paul Kennedy, Daniel Catchpoole and Nicholas Ho
  Functional Visualisation of Genes using Singular Value Decomposition
• Md. Sazzad Hussain, Hamed Monkaresi and Rafael A. Calvo
  Combining Classifiers in Multimodal Affect Detection
• Peter Sunehag, Wen Shao and Marcus Hutter
  Coding of Non-Stationary Sources as a Foundation for Detecting Change Points and Outliers in Binary Time-Series
• Qinxue Meng and Paul Kennedy
  Using network evolution theory and singular value decomposition method to improve accuracy of link prediction in social networks
• Kaimin Yu, Zhe Li, Genliang Guan, Zhiyong Wang and David Feng
  Unsupervised Text Segmentation using LDA and MCMC
• Helen Giggins and Ljiljana Brankovic
  VICUS - A Noise Addition Technique for Categorical Data
• Satya Gautam Vadlamudi, Partha Pratim Chakrabarti and Sudeshna Sarkar
  Anytime Algorithms for Mining Groups with Maximum Coverage
• Jialing Li, Li Li and Xiao Wen
  A Collaborative Filtering Recommendation System Combining Semantics and Bayesian Reasoning
• Omar S. Soliman and Amr Adly
  Associative Classification using a Bio-Inspired Algorithm

For further information contact AusDM@togaware.com