Keynote Speeches

Professor Ajith Abraham

In this talk, we present the importance of nature inspired meta-heuristic techniques for solving some of the complex global optimization problems. We start with the review of some of the popular algorithms based on Evolution and Swarm Intelligence and then we focus on some of the recent heuristics based on Foraging (Bacterial foraging optimization) and Music (Harmony search). Using Empirical studies, we illustrate the performance of these algorithms for solving global optimization problems.

Ajith's research and development experience includes nearly 20 years in the Industry and Academia. He works in a multi-disciplinary environment involving machine intelligence, network security, sensor networks, e-commerce, Web intelligence, Web services, computational grids, data mining and applied to various real world problems. He has given more than 40 plenary lectures and conference tutorials in these areas. He has published over 600+ publications and some of the works have also won best paper awards at International conferences and also received several citations. He Co-Chairs the IEEE SMC Technical Committee on Soft Computing. Currently he is also coordinating the activities of the Machine Intelligence Research Labs (MIR Labs), International Scientific Network of Excellence, which has members from over 60 countries. He has a world wide academic experience with formal appointments in Monash University, Australia; Oklahoma State University, USA; Chung-Ang University, Seoul; Jinan University, China; Rovira i Virgili University, Spain; Dalian Maritime University, China; Yonsei University, Seoul and Open University of Catalonia, Spain, National Institute of Applied Sciences (INSA), France and Norwegian University of Science and Technology (NTNU), Norway. For about 2.5 years, he was working under the Institute of Information Technology Advancement (IITA) Professorship Program funded by the South Korean Government. He received Ph.D. degree in Computer Science from Monash University, Australia and a Master of Science degree from Nanyang Technological University, Singapore. He serves the editorial board of several reputed International journals and has also guest edited over 35 special issues on various topics. He is actively involved in the Hybrid Intelligent Systems (HIS); Intelligent Systems Design and Applications (ISDA); Information Assurance and Security (IAS); and Next Generation Web Services Practices (NWeSP) series of International conferences, besides other conferences. He is a Senior Member of IEEE , IEEE Systems Man and Cybernetics Society, IEEE Computer Society, IET (UK), IEAust (Australia) etc. More information at: http://www.softcomputing.net

Professor Wen-Lian Hsu

In natural language processing, many machine learning models, such as maximum entropy, conditional random fields, are based on feature selection and parameter tuning. If the test set is not sufficiently correlated with the training set, the model is less likely to perform well. However, in practice, it looks as though the training set is never large enough to cover most test cases, especially those in user domains. Although adaptation would usually ease the problem a little, current machine learning models do not seem to catch the semantics or the context in free texts, which makes them less portable. After all these years, the following questions remain unanswered: "What kind of knowledge should be extracted from a training set to make the model more general?" "How do we minimize the laborious human annotation task on the training set?" "How do we combine knowledge from heterogeneous sources effectively?" "Which approach is better: knowledge-based or machine learning?" In this talk we shall discuss strategies that attempt to tackle these challenges.

Wen-Lian Hsu received a B.S. in Mathematics from National Taiwan University, a Ph.D. in operations research from Cornell University. His earlier work was on graph algorithms. He then applied similar techniques to tackle computational problems in biology and natural language.

Dr. Hsu is the inventor of the popular Chinese Input Method GOING, in 1993, which is now used by over 1 million users daily in Taiwan. Dr. Hsu's lab has won many international contests in natural language systems, including: the 1st place in the NTCIR6 2007 CLQA Chinese Question Answering (QA) Contest; the 1st place in NTCIR6 2007 CLQA English-to-Chinese Cross-Lingual QA Contest; the 1st place in NTCIR5 2005 CLQA Chinese QA Contest; the 1st place in SIGHAN 2006 Word Segmentation Contest; the 2nd place in SIGHAN 2006 Named Entity Recognition Contest; the first place in the BioCreAtIvE II.5 Interactor Normalization Task Challenge.

Dr. Hsu is particularly interested in applying natural language processing techniques to the understanding of DNA sequences as well as protein sequences, structures and functions and also to biological literature mining. He received many awards from the National Science Council, including the Distinguished Research Award in 1991, 1994, 1996 and the Appointed Distinguished Research fellow Award in 2005. He received the first K. T. Li Research Breakthrough Award in 1999, the IEEE Fellow in 2006, the Teco Technology Award in 2008, and the Outstanding Research Award of Pan Wen Yuan Foundation. He has been the president of the Artificial Intelligence Society in Taiwan from 2001 to 2002 and is currently the Director of the TIGP Bioinformatics Program in Academia Sinica.

Professor Jianrong Tan

Product customization has been recognized as an effective means to implement mass cus-tomization (MC). A new theory and method for MC-oriented evolutionary design of configuration product is presented based on the study of developing law of evolutionary design in integrated envi-ronment, which focuses on the innovation and reuse properties of configuration product.

The key technologies for general requirement modeling in quick response to customer requirement, multi-level stepwise configuration optimization driven by customer requirement and evolutionary deduction of product variable structure based on configuration association are thoroughly investigated. The suc-cessful application of the presented method in the development of real-life products demonstrates its utility, flexibility and robusticity.

Professor David Powers

In this talk we discuss a thirty year research focus on evolutionary and unsupervised learning techniques for the development of grounded natural language learning agent.
The research program has demonstrated algorithms that can discover features at the levels of speech, phonology, morphology, syntax, semantics and ontology, and also considers the evolution of language and the development of the mechanisms involved in language and ontology learning.
A surprising conclusion is that an individual language learner does not so much learn his mother's language as evolve a distinct idiolect optimized across a broader range of social contexts including co-evolution of peer culture and dialects.
The grounded learning approach has employed both simulated and physical robots, and also explores sensor fusion for speech and emotion recognition. Recently the technology has been turned around so as to act as a grounded Teaching Head in a hybrid virtual world, with a long term focus on completing the feedback circle and co-evolving the recognition and generation mechanisms for linguistic, social and emotional expression.
A critical motto of the research team is "one man's noise is another man's signal" as we address communicative behaviour as a coherent whole and broaden our scope from text to speech, gesture and emotion, and generalize from text interfaces to graphical, speech and brain control interfaces, address human factors relating to cognitive and perceptual load and conscious and unconscious aspects of language, behaviour and awareness.
This technology has featured in the MAGICIAN entry in the MAGIC Robotics Grand Challenge, in which the MAGICIAN team lead by Professor Powers is a finalist.

Prof. David M W Powers is Director of the Artificial Intelligence and Knowledge Discovery Laboratories at the Flinders University of South Australia, and program leader of the programs in Assistive Technology, Artificial Intelligence and Language Technology.
He has a Bachelors degree in Mathematics and Computer Science from the University of Sydney (1979) and a PhD in the area of Computational Psycholinguistics (1989) as well as other qualifications in Linguistics, Theology and Technical Analysis.
As an instigator of the field of Machine Learning of Natural Language and Ontology or Computational Psycholinguistics, Prof. Powers organized many events in this area, including founding ACL SIGNLL and CoNLL, the society and conference in Natural Language Learning.
Powers has a particular focus on biologically plausible and physically grounded unsupervised approaches to learning, as well as cross-supervised multimodal, adaptive, evolutionary and model fusion approaches.
Prof. Powers was also a pioneer in Parallel Algorithms, Concurrent Logic Programming and Bioinformatics, and currently holds contracts and grants relating to Brain Computer Computer Interface, Cognitive Neuroscience, Embodied Conversational Agents, Educational Technology, Robot Teaming, Human Factors and Human Computer Interface, some of which involve international collaboration with countries such as China, Germany and the US.
He is author or coauthor of a well known monograph on Machine Learning of Natural Language (Springer 1989) as well as around 200 publications including several patents. Dr Powers is also an advisory board member for multiple companies and has developed several technologies that are being exploited as start-up companies or under license to a major company (including YourAmigo and Clipsal Homespeak).
He is a member of several professional societies, and a senior member of IEEE.