skip to: page content | links on this page | site navigation | footer (site information)
ICONIP 2008 - 15th International Conference on Neural Information Processing of the Asia-Pacific Neural Network Assembly
November 25-28, 2008, Auckland, New Zealand
Special Sessions

Data Mining Methods for Cybersecurity

Organizers:

Prof. Youki Kadobayashi (Graduate School of Information Science, Nara Institute of Science and Technology, Nara Institute of Science and Technology, Japan)

Dr. Daisuke Inoue, Senior Researcher, Network Security Incident Response Group, Information Security Research Center, National Institute of Information and Communications Technology (NICT)

Dr. Tao Ban ( Information Security Research Center, National Institute of Information and Communications Technology, Japan,)
Contact: youki-k@is.aist-nara.ac.jp, bantao@nict.go.jp, Webpage: http://www.nict.go.jp

Aims and Topics:
Upon the explosive growth of computer and network communications in our daily life, people have become more than ever cared about the security of the information stored in their computers and being transmitted through the networks. Cybersecurity is raised to discuss the methodology and strategy that we can use to protect our data and resources from threats to confidentiality, integrity, and availability, in particular to safeguard our systems and infrastructure with an aim to instill confidence in online trade, commerce, banking, telemedicine, e-governance, and a host of other applications. The research of Cybersecurity has gained widespread public awareness over the past decades, now has been taken one part of national security, and wears the more importance from the government of most countries.
Recent researches on Cybersecurity have identified that Machine Learning and Data Mining (ML/DM) is a promising and reliable counterpart of the traditional signature-based techniques for protecting servers and PCs from malicious software, endless spam, cunning attackers, and persevering intruders. However, adapting classical ML methods and designing new DM methods to fulfill the exorbitant expectation of end users and to keep up with the advances of network infrastructure and hardware, it is just starting point of problem-solving for Cybersecurity. Compared to the traditional learning problems that ML methods were originally derived from, contemporary Cybersecurity problems exhibit the following different features: huge amount of information usually presented as data streams, real-time response requirement to prevent fast-spreading threats, dynamic characteristics of data sources, severe sampling bias in the training data, and inequality of misclassification costs. This poses a new critical challenge to the computer science society, especially to network security engineers and researchers from the field of computational intelligence.
The aim of this special session on Cybersecurity is to bring together scientists from the discipline of computational intelligence and researchers from network infrastructure and application sciences to demonstrate state-of-the-art Cybersecurity systems and to address the problem of how data mining can be further employed to fight against future cyber threats.

Topics of interest include but are not limited to:

  • Data mining for intrusion detection;
  • Cyber forensic analysis;
  • Spam filtering algorithms;
  • Privacy-preserving data mining;
  • Fraud analysis.
  • Fast Data mining over Huge volume data

Intelligent Video Surveillance and Analysis

Organisers:
Weng Kin LAI (lai@mimos.my), Centre for Advanced Informatics, MIMOS Berhad, Malaysia
Chee Peng LIM (cplim@eng.usm.my), School of Electrical & Electronic Engineering, Universiti Sains Malaysia, Malaysia

Aims and Topics:
With the proliferation of inexpensive but high quality cameras, it has become economically and technically feasible to employ multiple cameras to provide continuous round-the-clock surveillance. Furthermore, by automatically analysing human behavior and detecting the presence and/or absence of various objects in real time, intelligent video surveillance promises enhanced security at an overall lower cost while delivering high levels of efficiencies. However, a key challenge is to handle an immense flood of information coming from such systems when deployed in public areas. Moreover, visual inspection for even a small percentage of this flood of information for suspicious behavior that might betray a sinister motive poses various technical challenges.

This special session will provide an international forum for researchers and academicians to address some of these challenges, including but not limited to adaptive intelligent systems, intelligent sensors, sensor fusion, gait analysis, motion analysis, behaviour modeling, camera control, low-level feature extraction, and intelligent multimedia signal processing.

Computational Models and their Applications in Machine Learning and Pattern Recognition

Organizers:
Dr. Kazunori Iwata
Hiroshima City University
kiwata(at)hiroshima-cu.ac.jp
http://www.robotics.im.hiroshima-cu.ac.jp/~kiwata/

Prof. Kazushi Ikeda
Nara Institute of Science and Technology
kazushi(at)is.naist.jp
http://hawaii.naist.jp/~kazushi/index-e.htm

Aims and Topics:
Most of algorithms in machine learning (ML) and pattern recognition (PR) are based on computational models. For example, reinforcement learning in ML is based on a model of interactions between an agent and an environment, and its model is usually formulated as a Markov decision process. Another example is video processing in PR, and the time-series model of video frames is often represented as a hidden Markov model. Thus, computational models play an important role in their applications. Accordingly, for applying computational models to ML and PR, it is very fruitful to clarify and/or organize mathematical properties satisfied in computational models under some constraint.
This session focuses on mathematical analysis of well-known computational models and their neat applications in ML and PR. Bringing a new computational model into ML and PR is also welcome. The aim of this session consists in providing the participants an up-to-date account on recent theoretical developments of computational models. We encourage submissions on one or more of the following topics related to ML and PR.

  • Mathematical Analysis of Computational Models
  • Organization of Computational Models from a New Point of View
  • Proposition of New Computational Models
  • Applications Based on Computational Models

Lifelong Incremental Learning for Intelligent Systems

Organizers:
Dr. Seiichi Ozawa (Kobe University, Japan)
ozawasei@kobe-u.ac.jp

Dr. Paul Pang (Auckland University of Technology, New Zealand)
spang@aut.ac.nz

Dr. Minho Lee (Kyungpook National University, Korea)
mholee@knu.ac.kr

Dr. Guang-Bin Huang (Nanyang Technological University, Singapore)
EGBHuang@ntu.edu.sg

Aims and Topics:
Lifelong incremental learning provides a general learning paradigm to build fully automated and autonomous intelligent systems that can improve the performance not only for a single task but also for a series of multiple related tasks on their own. Lifelong incremental learning includes several crucial learning schemes such as on-line learning, incremental learning, autonomous learning, and multi-task learning, each of which is very fascinating to those who attempt to develop future intelligent systems for pattern classification, forecasting, robotics, diagnosis, etc.
The aim of this special session is to provide an opportunity to share the latest results on lifelong incremental learning which could range from theoretical approaches to building intelligent systems.

Topics of interest include but are not limited to:

  • Theoretical approach to lifelong incremental learning
  • Supervised lifelong incremental learning
  • Unsupervised lifelong incremental learning
  • Semi-supervised lifelong incremental learning
  • Multi-task learning
  • On-line feature extraction
  • Development of real time intelligent system

Application of Intelligent Methods in Ecological Informatics

Organisers:
Dr Michael J. Watts
michael.watts@bio.usyd.edu.au

Dr Susan P. Worner
worner@lincoln.ac.nz
http://www.lincoln.ac.nz/story1666.html

Aims and Topics:
In recent years there has been growing awareness amongst ecologists of the value of machine learning and ecological informatics techniques for the analysis of ecological data and knowledge discovery and prediction. Ecological data is often complex, noisy and sparse and presents certain challenges to the data analyst and ecological modeller. Intelligent methods are increasingly being used to analyse and model such data as shown by the relatively large number of papers in the ecological modelling literature where there have been at least sixty articles in the last decade using techniques like ANN. The area of neurocomputing is constantly changing as new approaches are developed and new theory emerges. This session has the potential to bring together researchers from diverse disciplines such as the biological and ecological sciences and researchers in intelligent techniques, neurocomputing and knowledge discovery.

The proposed special session will cover the application of intelligent methods such as artificial neural networks (ANN), fuzzy systems (FS) and evolutionary algorithms (EA) to the analysis of ecological data (ecological informatics). Ecological informatics is concerned with the integration and processing of ecological data using biologically inspired computing (or neurocomputing) such that ecological data is refined into ecological information, ecosystem theory and ecological decision support (Recknagel, 2002).

Topics of interest include, but are not limited to, the following:

  • The use of intelligent methods in modelling species distributions
  • Data mining of ecological data
  • Analysis of species assemblages with intelligent methods
  • Fusions of ANN, FS, EA or other intelligent methods applied to ecological data
  • Processing of ecological data from spatial information systems using intelligent methods
  • Issues in the preparation of ecological data for modelling with intelligent methods
  • Uncertainty and intelligent methods

Pattern Recognition from Real-world Information by SVM and other Sophisticated Techniques

Organizers:

Prof. Ikuko Nishikawa
Ritsumeikan University
nishi[at]ci.ritsumei.ac.jp
http://research-db.ritsumei.ac.jp/Profiles/36/0003516/profile.html

Prof. Kazushi Ikeda
Nara Institute of Science and Technology
kazushi[at]is.naist.jp
http://hawaii.naist.jp/~kazushi/index-e.htm

Aims and Topics:
Neural networks have been a standard learning machine for pattern recognition, and as a new trend other sophisticated techniques have attracted much attention in a decade, such as support vector machines (SVMs), committee machines and graphical models.  Recent techniques are discussed in the Bayesian statistical framework.
This special session focuses on the recent and sophisticated techniques for the pattern recognition, from the modeling to the applications aiming to extract the real-world information. The topics of interest are not limited to the above methods but also include other related techniques such as independent component analysis (ICA) for blind source separation, hidden Markov models (HMM) for time series modeling of the real-world data, to bring together the researchers in these fields and exchange the new ideas, trends, current research results and the application issues.  

Topics of interest include the pattern recognition by

  • SVM and other kernel methods
  • Committee models and ensemble learning such as boosting and bagging
  • Blind source separation by Independent Component Analysis and other methods
  • Statistical time series models such as hidden Markov models
  • Text analysis by N-gram and other methods

Dynamics of Neural Networks

Organizers:

Dr. Zhigang Zeng
School of Automation, Wuhan University of Technology
Wuhan, 430070, China
zgzeng@gmail.com

Dr. Tingwen Huang
Texas A&M University at Qatar
Doha, P.O.Box 5825, Qatar
tingwen.huang@qatar.tamu.edu

Aims and Topics:
In the past two decades, the dynamics of neural networks, especially the stability of neural networks, have been extensively investigated because of their important applications in various areas such as pattern recognition and combinatorial optimization. So far, we still have a lot of challenges in discovering easily verified criteria to guarantee the neural networks being stable and studying other dynamics property of neural networks.
The aim of this special session is to provide a forum to share the latest results on dynamics of neural networks, fuzzy neural networks and complex networks.

Topics of interest include but are not limited to:

  • Stability of neural networks
  • Bifurcation and Chaos of neural networks
  • Mathematical modeling of neural networks
  • Neural dynamical analysis

Recent Advances in Brain-inspired Technologies for Robotics

Organizers:

Kazuo Ishii (ishii@brain.kyutech.ac.jp), Kyushu Institute of Technology, Japan
Keiichi Horio (horio@brain.kyutech.ac.jp), Kyushu Institute of Technology, Japan

Aims and Topics:
Recently brain-inspired technology has been taking an important role to cope with problems, pattern recognition, decision making or navigation in a real world. Especially, in robotics, many serious problems remain to realize intelligent robots acting a real world. Robotics research was initiated by industrial applications as typified by manipulators aiming at industrial process automation. In parallel with the advancement of processing hardware and software, robotic applications have broadened from industry to service areas, such as, guidance, patrol, tele-operation, surgery, rehabilitation, entertainment, Robocup, etc. In order to realize new generations of robots that can carry out these tasks reliably and efficiently, several technical issues remain open, involving work on sensory systems, actuators, recognition, learning, behavior planning, control, etc. Brain-inspired technologies can overcome the limitations of conventional approaches in many engineering problems and are also expected to lead to breakthroughs in robotics research. In this special session, we invite contributions on brain-inspired technologies for robotics. The subject areas include but are not limited to:

  • Brain-inspired recognition system
  • Brain-inspired decision making
  • Brain inspired SLAM
  • Neural robotics, cognitive robotics, developmental robotics
  • Autonomous intelligent systems capable of real-world interaction
  • CPG, neural oscillator applications to robotics
  • Neuromorphic visual, auditory, and tactile sensors.
  • Other brain-inspired robotics applications

Neural Information Processing in cooperative multi-robot systems

Organizers:

José A. Becerra
Integrated Group for Engineering Research, Universidade da Coruña
ronin@udc.es
http://www.gii.udc.es/jose_becerra

Javier de Lope
Perception for Computers and Robots, Universidad Politécnica de Madrid
jdlope@fi.upm.es
http://www.dia.fi.upm.es/~jdlope

Iván Villaverde
Computational Intelligence Group, Universidad del País Vasco
ivan.villaverde@ehu.es
http://www.ehu.es/ccwintco/index.php/Usuario:Ivan.villaverde

Aims and Topics:
Multi-robot systems are emerging as a new frontier in Robotics research, posing new challenges and offering new solutions to old problems. Multi-robot systems are usually employed to solve tasks that are impossible, or very difficult, for just one robot. They are also used to solve, in a more efficiently way, tasks that can be realized by isolated robots. Finally, multi-robots systems may provide a means to introduce redundancy in critical or dangerous tasks, where the possibility of losing a robot is high, or in situations where information is fragmented or incomplete. Multi-robot systems may be considered a particular case of multi-agent systems, so they are collections of interacting and cooperating autonomous agents, in this case with physical embodiment that impose restrictions on what they can do and how they interact with the environment.
The aim of this special session is to present state of the art papers on multi-robot systems that make use of Artificial Neural Networks and other biological inspired techniques like Evolutionary Algorithms, which address current problems in this research area from a biological inspired point of view. Thus, the topics of interest include:

  • Heterogeneous and modular robot architectures.
  • Heterogeneous multi-robot team coordination.
  • Cooperation hierarchies.
  • Cooperative multi-robot exploration.
  • Inter-robot communications.
  • Reactive distributed behavioural teams.
  • Perception based control: visual servoing and SLAM.
  • Self-organizing robot teams.
  • Multi-agent strategies for multi-robot systems.


©2008 Knowledge Engineering and Discovery Research Institute (KEDRI)