|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
2003 Seminar SeriesSee below for detailed abstracts
Abstracts Action Planning: Recent Theoretical and Practical Advances Abstract: Action planning is one of the challenging research areas in Artificial Intelligence. In the previous decade, a number of new algorithmic approaches to planning have been developed and some of the theoretical problems have been solved. After a brief introduction to action planning, I will first sketch the range of planning frameworks and formalisms, which are currently in use, and will then go on and sketch a new planning technique that has enabled us to solve much bigger problems -- if one considers the benchmark problems from the literature. One of the interesting questions is then, whether and how the technique can be extended to more powerful planning formalisms. In order to answer this question, I will introduce a formal framework for translating between different planning formalisms and report on results that have been achieved in this framework. In particular, I will talk about some recent results concerning domain axioms in planning languages.
About the speaker:
Bernhard Nebel received his Ph.D. (Dr. rer. nat.) from the University
of Saarland in 1989. Since 1996 he has been Professor at
Albert-Ludwigs-University Freiburg and head of the research group on
foundations of AI. His research interests are in knowledge
representation, planning, and robotics. He is the author of more than
90 refereed scientific papers. Among other professional services he
has been program chair of the International Joint Conference on AI
(IJCAI) in 2001 and he is Research Note editor of the Artificial
Intelligence journal. In 2001, Bernhard Nebel was elected as an ECCAI
fellow. See http://www.informatik.uni-freiburg.de/~nebel/ for more
details.
Multi-version Concurrency Control to Improve Data Availability in Mobile Computing Abstract: Mobile-computing environment is characterized by limited execution capability at mobile hosts, low bandwidth and the relatively high costs of wireless connection, and frequent disconnections and mobility of mobile hosts. A transaction in mobile computing environment is different from the transactions in centralized or distributed databases. In general, to support mobile computing, the transaction processing models should accommodate its limitations. I will present in-depth performance results of a multi-version transaction (MV-T) model, which exploits versions to increase availability in mobile databases. Each transaction is either in start, committed or terminated state. A transaction can start and commit at mobile host (MH) but terminates only at mobile service station (MSS). We present performance of a two-version model, where each data object can have two versions, one committed and the other is terminated. A read transaction is never blocked, as it is always made available either a committed or a terminated version. The model is a hybrid model which uses both locking and timestamps and is deadlock free. Our model increases read-write and write-write availability and abort-rate decreases for high write-probabilities with moderate load. The blocked transactions have also shown low restart-rate. Therefore, the model is very efficient in a mobile computing environment where low abort-rates are required and at the same time, blocking does not result in high restart-rate.
About the speaker:
Dr Sanjay Kumar Madria received his Ph.D. in Computer Science from
Indian Institute of Technology, Delhi, India in 1995. He is an
Assistant Professor, Department of Computer Science, at University of
Missouri-Rolla, USA. He has published widely in reputed journals and
conferences in the areas of web warehousing, mobile databases, nested
transaction management and performance issues. He guest-edited WWW
Journal and Data and Knowledge Engineering Sp. Issues on Web data
management and Data warehousing. He has served as Program Chair for
International conferences and workshops. He is PC member of various
database conferences and serves as reviewer for many reputed database
journals such as IEEE TKDE, IEEE Computer, ACM Internet Computing
etc. He serves as invited panelist in NSF and Sweden Council of
Research for selection of research grants. He was invited keynote
speaker in Annual Computing Congress in Oct.99 in Canada and invited
speaker in International Conference on Information Technology,
2001. He is an IEEE Senior Member.
Some combinations of Fuzzy Logic and Neural Network Abstract: Networks and Fuzzy Logic are the most important techniques for real applications in the field of "softcomputing". Both have their advantages and disadvantages. After a short introduction in both theories we present 3 different combinations for a successful application. - prediction of the maximal error of a neural network by the help of fuzzification - optimizing the parameters of a learning rule for neural networks by the help of a fuzzy controller - optimizing fuzzy rules based systems (eg fuzzy controllers) by the held of neural networks
About the speaker:
Wolfram Lippe is a Professsor in the Department of Computer Science at
the Wesfliche Wilhelms University of Munster.
Hybrid Prefix Codes for Practical Use Abstract: Minimum-redundancy prefix codes are widely used, and approximate codes have also received considerable attention. Flat codes, which may be considered as a class of approximate codes, have an extremely simple structure which facilitates fast decoding, but their compression effectiveness is often poor. Minimum-redundancy prefix codes, on the other hand, are slower to decode than flat codes, but minimize the compressed file size. We present a hybrid code which exhibits the simple structure of a flat code and retains much of the compression effectiveness of a minimum-redundancy code. The structural constraints of the hybrid code should enable fast decoding and also provide for fast compressed string searching. We discuss properties of the hybrid codes and a method for their efficient calculation.
About the speaker:
Alistair Moffat has been at the University of Melbourne,
Australia, since 1986 as a teacher and a researcher. He is the
co-author of two books on compression and information retrieval
("Managing Gigabytes: Compressing and Indexing Documents and Images,
Morgan Kaufmann, second edition, 1999; and "Compression and Coding
Algorithms", Kluwer, 2002), and more than one hundred refereed
research papers. His most recent project is an introductory text
"Programming, Problem Solving, and Abstraction with C"), published by
Pearson SprintPrint in early 2003.
Cost-Effective Development and Delivery of 100% Online IT Courses Abstract: The pedagogical effectiveness of internet technology for online education is still not well understood (Hartley, 1996). How is learning maximised when teachers and students are connected only by a learning management system (LMS) while separated by distance, and possibly time? Previously, we argued the focus should be on what students do, rather than what teachers do (Fernandez, 2001; Fernandez, John & Netherwood, 2001). Thus, we coordinate subject material and activities around learning objectives, targeted at achieving clearly identified learning levels and stated learning outcomes. Here we continue this approach while addressing the cost-effective development and delivery of 100% online courses in Information Technology (IT). Our aim is to maximise flexible learning opportunities for off-campus, distance students, cost-effectively. We utilise a palette of WWW tools to provide subject content, self-test questions, interaction, assignments, and assessment feedback. This project has produced twenty 100% online IT courses in three years, reaching 6000 off-campus students, world-wide, via freely downloadable WWW browser and associated software (Zuluaga, 2002a). Based on these online courses, a complete Bachelor degree in IT is offered through Open Learning Australia (OLA Handbook, 2002). The first graduates were awarded their degrees in 2001.
About the speakers:
Catherine Zuluaga, Ed Morris and George Fernandez are respectively a
Senior Lecturer, Senior Lecturer and Associate Professor in the School
of Computer Science and Information Technology at RMIT.
Linear time prediction of protein structures Abstract: Most known techniques for modelling of proteins base the prediction either on calculations of molecular forces or on finding single best matches of complete strings. We determine sets of short and perfect matches of sub-strings. From these sets we extract statistical information related to the structure including confidence indicators on which we base the prediction of the complete structure. While current ab initio structural prediction algorithms for proteins need substantial supercomputing time, we propose new algorithms that run in linear time thus making structural prediction of proteins available to any researcher equipped with a PC.
About the speaker:
Heiko Schroder is Professor and Head of the School of Computer Science
and Information Technology at RMIT University.
Computational Models for Goal-Driven Systems Abstract: An automated space probe is about to land on Mars. Its resources are limited, the environment is constantly changing and it has to make decisions quickly. Clearly it is impossible to anticipate all possible contingencies, and so it is necessary to specify the behaviour of the probe by the goals it is trying to achieve rather than as a pre-determined sequence of actions. In this talk we will describe how such systems can be designed. In particular, we concentrate on a particular computational interpretation of goals, and show how this can realised in a framework based on linear logic.
About the speaker:
James Harland is a Senior Lecturer in the School of Computer Science
and Information Technology at RMIT University.
Matrix Decompositions for Mining Complex Data Abstract: The values in real-world datasets often result from the complex interactions of a number of different processes. For example, observations of astronomical objects reflect the properties of the objects themselves, the effects of the atmosphere, and properties of the observing devices. Many mainstream data mining techniques perform poorly on such datasets. Matrix decompositions, on the other hand, seem to be effective at separating the processes that generate complex data, allowing each to be analysed separately. Two decompositions in particular, singular value decomposition and semidiscrete decomposition, have been used. I will describe how they work, and the kind of results that have been obtained in applications such as astrophysics, geochemistry, and intrusion detection.
About the speaker:
David Skillicorn is a professor in the School of Computing at Queen's
University in Canada. His work has focused on the role of data in computation,
particularly parallel and high-performance computation. He was involved in
the development of one of the first parallel computing models, in which the
same program could run unaltered on different kinds of parallel computers;
and in the design and implementation of BSP, which was instrumental in
showing the real cost structure of parallel computers and clusters.
He has worked in data mining for the past 6 years, developing new sequential
and parallel techniques. At present his focus is on understanding complex
datasets in applications such as astrophysics, geochemistry, network
intrusion, fraud detection, and counterterrorism. He has an undergraduate
degree from the University of Sydney and a Ph.D. from the University of
Manitoba. He is on sabbatical at UTS in Sydney until June 2003.
Oh what a tangled web - the network model of complexity Abstract: In today's global society, one of the greatest challenges is to cope with the complexity that results from the intricate dynamically changing networks that link individuals, societies, economies, and environments. This overview touches on some of the theoretical, practical and applied problems that arise. Key theoretical ideas, such as the network model and natural computation, also have many applications to environmental, biological and social systems. In this talk, I will provide examples drawn from my experiences in research, government and international agencies.
About the speaker:
David Green is a professor in the School of Computer Science and
Software Engineering at Monash University. In a research career
spanning more than thirty years, he has applied computers to problems
as diverse as starfish, bushfires, DNA, and social issues. He is
editor of the journal Complexity International and author of several
books on the new field of complexity. His research has also included
online GIS and distributed information systems.In the early 1990s, he
established several pioneering online information services, such as
the Guide to Australia and the New South Wales HSC Online. He also
played a key role in international efforts to compile comprehensive
databases of the world's biodiversity.
Primality Testing and its Importance for Cryptography Abstract: The introduction of private key cryptography in the 1970s and its initial development based on the factorization of numbers led to much work on primality testing. We examine some of the key algorithms in this area, concluding with the unexpected polynomial time algorithm discovered by Agrawal, Kawal and Saxena in August 2002.
About the speaker:
Lynn Batten is Professor of Mathematics at Deakin University and holds
the Deakin Chair in Mathematics. She has worked in combinatorial
geometry for many years and is particularly interested in applications
of this area to communications security and computational chemistry.
Before arriving in Australia in January 2000, she was Head of the
Department of Mathematics at the University of Manitoba (1989 to 1996)
and then Associate Dean of Academic and Industrial Research in the
Faculty of Science.
DLS, WebLearn, Blackboard etc. Abstract: RMIT provides staff with a central facility to better manage teaching and learning, supported by the Distributed Learning Systems (DLS) group. This has been implemented as the Learning Hub, a university-wide system that provides students with a unified interface to access subject material and a palette of learning tools. One of those tools is WebLearn, a Web-based, student-centred learning environment developed by Computer Science that has been successfully used by RMIT lecturers over the last few years. This talk will introduce the facilities provided by the Learning Hub and the supporting DLS team. It will also briefly introduce WebLearn for new staff, and discuss the fundamental concepts of the system and how it may be used to teach Computer Science. The talk will also present the new features of the WebLearn system, including the Maple interface, the new WebTutor developments related to adaptive learning, and some ideas about the possibilities of the system for CS&IT.
About the speaker:
George Fernandez is an Associate Professor in the School of Computer
Science and Information Technology at RMIT University.
Abstract: Over the last decade a number of Component Based Architectures (CBAs) have been developed to facilitate language interoperability. The most widely known and used of these would be the Object Management Group's (OMG's) Common Object Request Broker Architecture (CORBA) and Microsoft's Component Object Model (COM) and Distributed COM (DCOM). Although not specifically designed to provide language interoperability, many Java based technologies, such as Enterprise Java Beans (EJB), also adhere to the principles of a CBA. The Common Language Infrastructure (CLI) is the latest CBA to be developed and builds on many of the lessons learnt from previous architectures. In this seminar we will look at the major components of the CLI. The CLI is the ECMA standard on which Microsoft has built three implementations: the .NET Framework, the Compact Framework and the Shard Source CLI. The CLI is also the specification on which non-Microsoft implementations, such as Ximian's Mono, are based. The first hour of this talk will described the CLI specification and the second hour will describe implementations.
About the speaker: Damien Watkins is the founder and Managing Director of Project 42 P/L and a Research Associate with the Distributed Systems Technology Centre. Prior to commencing Project 42 Damien lectured at the School of Computer Science and Software Engineering at Monash University. Damien lectured both postgraduate and undergraduate subjects, covering the areas of UNIX System Call Programming, Windows Programming with C++, Component Based Software Development and Distributed Object Oriented Technologies. Damien taught the world's first .NET Framework course at Monash University. He has also taught at Uppsala University (Sweden) and KMITNB (Thailand).
Since 1998 Damien has been engaged with Microsoft Research on the
development of the .NET Framework. Component architectures that provide
language interoperability have always held a keen interest for Damien,
and he has written a paper on this topic, titled 'Handling Language
Interoperability with the Microsoft .NET Framework.'
User Centred Design of Smart Internet Technologies Abstract: The Smart Internet Technology Cooperative Research Centre (SITCRC) involves 11 universities and five industry partners. It has four technology programs dealing with Smart Personal Agents, Natural Adaptive User Interface, Intelligent Environment and Smart Networks. The User Environment program intersects these technology programs to ensure that the technologies are useful and easy to use. In this presentation, I will speak about the challenges of communicating with the technology researchers at the discovery phase of UCD in an academic context. We found that we needed two-way communication, rather than a UCD person attached to a technology program. Personas and scenarios were useful communication tools at this early stage. It also helped to have champions of UCD within the technology programs. About the speaker:
Assoc Prof Supriya Singh leads the User-Centred Design project in the
Smart Internet Technology CRC. She is a Senior Research Fellow in the
Faculty of Business. Her research interests are in the use and design
of information and communication technologies; the sociology of money
and communications; and qualitative research. Supriya is the President
of the Association for Qualitative Research and an affiliate without
term of the Program on Information Resources Policy at Harvard
University. For more details see
www.rmit.edu.au/bus/rdu/supriya.
A Non-dominated Sorting Particle Swarm Optimizer for Multiobjective Optimization Abstract: In this talk, I will introduce a modified Particle Swarm Optimizer (PSO), Non-dominated Sorting Particle Swarm Optimizer (NSPSO), for better multiobjective optimization. PSO is an optimization technique inspired by studies of social behaviour of insects and animals, and has been found very effective for solving single objective optimization problems. NSPSO extends the basic form of PSO by making a better use of particles' personal bests and offspring for more effective non-domination comparisons. Instead of a single comparison between a particle's personal best and its offspring, NSPSO compares all particles' personal bests and their offspring in the entire population. This proves to be effective in providing an appropriate selection pressure to propel the swarm population towards the Pareto-optimal front. By using the non-dominated sorting concept and two parameter-free niching methods, NSPSO and its variants have shown remarkable performance against a set of well-known difficult test functions (ZDT series). Our results and comparison with NSGA II (one of the current best multiobjective optimizers) show that NSPSO is highly competitive with existing evolutionary and PSO multiobjective algorithms. A research paper on NSPSO has been accepted and nominated for a best paper award at GECCO'03, which will be held in Chicago, from 12-16 July. For more info, see: http://gal4.ge.uiuc.edu:8080/GECCO-2003/ About the speaker:
Xiaodong Li is a lecturer in the School of Computer Science and IT, RMIT
University. He received his Ph.D. in Artificial Intelligence from Otago
University, New Zealand, 1997.
Extending Agents by Transmitting Protocols in Open Systems Abstract: Agents in an open system communicate using interaction protocols. Suppose that we have a system of agents and that we want to add a new protocol that all (or some) agents should be able to understand. Clearly, modifying the source code for each agent implementation is not practical. A solution to this problem of upgrading an open system is to have a mechanism that allows agents to receive a description of an interaction protocol and use it. In this paper we propose a representation for protocols based on extending Petri nets. However, this is not enough: in an open system the source of a protocol may not be trusted and a protocol that is received may contain steps that are erroneous or that make confidential information public. We therefore also describe an analysis method that infers whether a protocol is safe. Finally, we give an execution model for extended Petri nets.
About the speaker:
Lavindra de Silva is a PhD student in the School of Computer Science
and Information Technology at RMIT.
Exploiting Regularities in Data for Bioinformatics Abstract: The research community is inundated with data such as the genome sequences of various organisms, microarray data and so on, of biological origin. This data-volume is rapidly increasing and the process of understanding the data is lagging behind the process of acquiring it. The sheer enormity calls for a systematic approach to understanding using computational methods. As a first step towards making sense out of the data, we study the regularities in the data and hypothesize that this reveals vital information towards greater understanding of biological systems. The talk will focus on various kinds of regularities in data, that we identify and devise methods for unsupervised (automatic) discovery. For genomic data that is one dimensional (genome or protein sequences), we identify substring and permutation patterns. More generally, we define 2D patterns, association patterns for microarray data and network motifs for metabolic pathways. We will give mathematical definitions of these problems, present some theoretical results as well as open problems. We will conclude the talk with various concrete examples of applications of motif discovery in the area of computational biology.
About the speaker:
Laxmi Parida has been a research staff member since 1998 at IBM
T.J. Watson Research Center in the Bioinformatics and Pattern
Discovery Group at the Computational Biology Center.
Abstract: In spite of the hype, it is clear that web services are likely to change the way we build loosely coupled distributed systems. This session starts with a vendor-neutral look at the vision and standard protocols behind web services. Following this, we'll drill into how .NET implements those standards, specifically looking at how web services can be built using the .NET base class libraries. Finally, we will look at how the web services standards are being enhanced going forward to include additional functionality (i.e. security).
About the speaker:
Nigel Watson is a staff member at Microsoft Australia.
Abstract: Apple Computer, traditionally a strongly innovative technology company, has seen some important changes in recent times. Mac OS X is a highly innovative Operating System, based on Unix, Open Source and Open Standards. It includes powerful, free programming and development tools, and is a very flexible development platform. Apple has also introduced a raft of innovative hardware, from exceptionally well designed laptops, to powerful desktops, and strategically importantly, new servers and storage products. This lunchtime talk will provide a brief overview of Apple's new software and hardware, and provide a forum to ask technical questions of an Apple Systems Engineer.
About the speaker:
Joseph Cox is a staff member at Apple Australia.
What computing should we be teaching to non-computing students? Abstract: How the times have changed! Children are learning word processing, spreadsheeting, Internet and email use at home and at high school or even primary school. No one talks about computer literacy now. Do we need to teach computing to non-computing students who are already computer literate? If yes, what should be taught? One approach is to follow the "Fluency in IT" approach of Larry Snyder. We discuss this approach and highlight its weaknesses and propose another, hopefully more interesting, approach to teach computing to non-computing students. About the speaker:
Gopal Gupta is Deputy Dean of the Faculty of Information Technology at Monash University.
The Semantic Web, Applications, and Migration at HP Labs Abstract: After a short introduction of Hewlett-Packard Laboratories, the motivating problem of Application Integration and its difficulties is described. The presentation highlights research issues in this domain and shows why and how semantic modeling is a disruptive technology. We then cover the state of the art in Semantic Web technology, describe the concepts and also more in detail the developments at HP Labs, in particular the - Jena, Joseki, Genesis and Nuin tools - leading the domain of the semantic Web from a standard towards the migration to industrial tools and large scale applications. About the speaker:
Dr Bernard Burg is the manager of the Associative Metadata
Dept at the HP Laboratories in Palo Alto. He leads research in the
development of large-size metadata Semantic Web stores. Prior to this
work, HP Labs have explored agent technology to prefigure the Internet
of tomorrow, whereby Web services better serve users by considering
context, including location and current user activities. Dr Burg is a
recognized expert in Agent technology where he was instrumental in
writing the FIPA standard, has realized the first agent platform on
phones (EU LEAP project) and has launched the large scale projects of
Agenticities, realizing the first open world-wide agent testbed.
Dynamic Load Balancing for Systems Under Heavy Traffic and High Task Variation* Abstract: * This project is supported by the ARC (Australian Research Council under Discovery scheme) and SUN Microsystems. The most critical property exhibited by a heavy-tailed workload distribution is that a very small fraction of the very largest tasks make up a large fraction of the workload. This property makes the load very difficult to distribute using conventional task assignment policies (such as Random and LLF). Size-based task assignment policies (e.g. SITA-E, SITA-V, TAGS) have been proposed to handle heavy-tailed workloads, but suffer from limitations due to their static nature and assumption of prior knowledge of a task's service requirement. In this talk I will be reviewing some of the work we did in the area of load balancing for servers under heaving traffic and high sizevariation. In our early work, we proposed LFF (Least Flow Time) approach that avoids to unbalance the load caused by very large tasks. LFF allocates tasks proportionally to the processing capability of participating servers by taking into account their remaining processing time. LFF-PRIORITY dynamically computes two priorities, namely task size and task size priorities, and put them in a priority based multi-section queue. Lately we proposed TAESR (Task Assignment based on Estimating the Service Requirement) as an extension of TAGS. TAESR improves performance under heavy-tailed workloads by using multiple queues at each host, denoting restarted tasks as priority tasks and fast-tracking them to their final destination via priority queues. We found that TAESR performs significantly better (by a factor of three in some cases) than existing approaches under realistic conditions of high task size variation. About the speaker:
Zahir Tari is an associate professor at RMIT University. He
is the Activity Unit leader at School of Computer Science & IT,
RMIT (Australia). He acted as the program committee chair as
well as general chair of several international conferences (e.g.
DOA/CoopIS/ODBASE 02 and 03, DOA 00 and 01, IFIP DS
11.3 on Database Security, IFIP 2.6 on Data Semantics).
He is the co-author of several books. Dr Tari also
published in several reputable journals (such as ACM Transactions
on Databases and IEEE Transactions on Parallel and Distributed
Systems). More details about Zahir and his team can be found at
http://www.cs.rmit.edu.au/eCDS
WebLearn plus Maple: A Useful Formative Tool for Teaching and Learning Abstract: In this presentation, the current level of interaction between WebLearn and Maple will be described. Sample WebLearn/Maple quiz questions, some currently used in a variety of Engineering courses, ranging from simple to more advanced will be demonstrated, and the underlying code described. The range of possible informative feedback that can be produced using the symbolic manipulative capabilities of Maple will also be discussed. About the speaker:
Gary Fitz-Gerald is an Associate Professor in the Department of
Mathematics and Statistics at RMIT.
Modelling and Imaging Bioelectric Sources in the Heart Abstract: Everybody knows the Electrocardiogram ECG. But how does the body produce these "bioelectric" signals? Today it is possible to make computer models of the heart starting with the ionchannels in the cell membrane, then make computer simulations of single myocardium cells, then model the interaction between millions of cells, build together a complete heart and end up with the electric potentials at the body surface. This way many types of arrhythmias can be understood and in future it will be possible to carry out an individual therapy planning based on these computer models. Imaging bioelectric sources is a new functional imaging modality that is aiming at depicting the electrophysiological properties of the heart based on the measuremment of Body Surface Potential Maps BSPM and/or the magnetic fields produced by the bioelectric currents (Magentocardiography). Unfortunately the reconstruction of the sources from the measured signals is an "ill-posed problem" - so intelligent algorithms and regularization methods have to be found and tested. The talk will show where we are and where we want to go in modelling and imaging of bioelectric sources for better diagnosis and therapy planning in cardiology. About the speaker:
Olaf Doessel is a professor in the Institute of Biomedical Engineering
at the University of Karlsruhe.
Simplifying the Development of Intelligent Agents: Concepts, Methodology, and Tools Abstract: A software agent can be thought of as an autonomous object with goals. Agents are attracting increasing attention as an approach for building software that is to be deployed in environments that are complex, dynamic, or where failure is possible. This talk will report on work done in the context of "SAC" (Simplifying Agent Concepts), an ARC-funded Linkage project with industrial partner Agent Oriented Software. The talk will focus in particular on an agent-oriented software engineering methodology called Prometheus that Lin Padgham and myself have developed, and on tool support for agent-oriented software engineering. This is joint work with many people including Lin Padgham and James Harland. About the speaker:
Michael Winikoff is a senior
lecturer in the school of computer science and information technology at RMIT
University. His research interests include Intelligent Software Agents, Agent
Reasoning, Agent-Oriented Software
Engineering, Multi-Agent Systems, Open Systems, Logic and Proof Theory, Logic
Programming, Linear Logic, Advanced Programming Languages, and Formal Methods.
Development Models and Intellectual Property: Are Openess and Commercialization Compatible? Abstract: The open source software development model has been part of the software ecosystem for 30 years. This development model has increasingly moved into commercial use and is now affecting the business practices of most major commercial software producers. Jason Matusow, Microsoft's Shared Source Manager, will examine the broader effects of OSS and the implications of its long-term trend towards commercialization. Matusow will also discuss how Microsoft is learning from OSS and why Microsoft is sharing source code with customers, partners, academics and governments worldwide under the Shared Source Initiative. About the speaker:
Jason Matusow As program manager of the Shared Source Initiative at Microsoft Corp., Jason Matusow is responsible for working with internal and external constituencies to coordinate Microsoft's global source licensing strategy. The Shared Source Initiative has established the company-wide policy and framework regarding the sharing of Microsoft's most valuable intellectual property assets including Windows, CE and .NET technologies. Matusow also consults with governments, corporations, academics and analysts globally on software intellectual property issues. Since joining Microsoft in 1996, Matusow has worked in a variety of positions including serving as a network infrastructure technical specialist, working with Microsoft's enterprise customers on design and deployment issues. He was also the founding member of the Microsoft Year 2000 core team, spearheading company Y2K efforts and representing Microsoft's Y2K strategy and preparedness to governments and corporations globally.
Before joining Microsoft, Matusow founded his own PC and networking
business and led the network consulting group for a Microsoft Solution
Provider. Matusow is a graduate of Boston University.
Abstract: In 1975 Parsons developed his dictionary of musical themes based on a simple contour representation. The motivation was that people with little training in music would be able to identify pieces of music. We decided to test whether people of various levels of musical skill could indeed make use of a text representation to describe a simple melody query. The results indicate that the task is beyond those who are unmusical, and that a scale numeric representation is easier than a contour one for those of moderate musical skill. Further, a common error when using the scale representation still yields a more accurate contour representation than if a user is asked to enter a contour query. We observed an average query length of about seven symbols for the retrieval task. About the speaker:
Sandra Uitdenbogerd is a lecturer in the School of Computer Science and Information Technology at RMIT.
Expectation Reasoning in Agent Systems Abstract: Although the view of agents as intentional systems has been dominant for many years, the connections between an agent's internal reasoning and its external environment have not been properly formalised. We have addressed this problem by introducing a formal language to describe this connection through the concepts of observation and expectation. In this talk, we present a specific expectation reasoning system based on the agent's observation system to formally explore further how an agent can use its past observations and current expectations to drive its future observations and alter its knowledge. About the speaker:
Binh Tran is a PhD student in the School of CS&IT at RMIT, working on
Expectation Logic with James Harland and Margaret Hamilton.
Abstract: TBA About the speaker:
Rex Black (Rex_Black@RexBlackConsulting.com) brings over two decades
of software and systems engineering experience to his role as
President and Principal Consultant of Rex Black Consulting Services,
Inc.
(http://www.RexBlackConsulting.com),
an international software and hardware testing and quality assurance
consultancy. He and his consulting associates help clients such as
Bank One, Cisco, Comverse, Dell, Hitachi, the US Department of
Defense, Schlumberger, Williams Communications, and others with
training, consulting, contract staffing, and on-site, outsource, and
off-shore execution of testing, test automation, and quality assurance
projects. His first book, Managing the Testing Process, published in
June 1999, has sold over 15,000 copies in six continents. In July
2002, Wiley published the updated and improved second
edition. Following on the heels of these successful books,
Addison-Wesley published Critical Testing Processes in July 2003. He
has also written numerous articles, presented papers, and given
keynote speeches at conferences and events around the world.
Towards a cooperative co-evolutionary model for multi-objective optimization Abstract: Evolutionary algorithms provide a distinct advantage to multi-objective optimization because they can generate a population of optimal solutions. Co-evolution builds upon nature inspired evolutionary algorithms through the evolution of one or more populations where a fitness dependency is implicitly or explicitly introduced, between populations. Until now there has been little work demonstrating how co-evolution can facilitate multi-objective evolutionary algorithms beyond the application of single-objective co-evolutionary techniques. The talk will introduce the audience to multi-objective optimization. We will discuss some of the questions we hope to address, and present a cooperative co-evolutionary model. The proposed model co-evolves sub-components of solutions, where each component contributes in some way to the optimality of the combined solution. About the speaker:
Antony Iorio is a postgraduate student in the School of Computer Science and Information Technology at RMIT.
Task Control and Event Monitoring in a System of Cooperating Information Agents Abstract: The InfoSleuth project was an applied R&D project at MCC (Microelectronics and Computer Technology Corporation), which investigated the use of Java-based distributed, cooperating software agents for information-gathering, discovery, and analysis, supported by distributed monitoring and integration of information at multiple levels of abstraction. The agent system design was driven by a number of 'real-world' applications in domains such as healthcare, bioinformatics, and business intelligence. In this talk I will describe the InfoSleuth architecture, with particular focus on its approach to task control and event monitoring. I will also discuss several related lessons and issues that emerged from the project. About the speaker: Amy Unruh is a newly appointed ARC Research Fellow at the University of Melbourne Department of Computer Science and Software Engineering. Until recently, she was an adjunct faculty member at the University of Texas, and worked for two different startups. Prior to that, she was a member of the InfoSleuth Distributed Agents Program at Micro-Electronics and Computer Technology Corporation (MCC), an industrial R&D lab. She has a Ph.D. in CS/AI from Stanford University, in the area of AI planning, and a BSc in Computer Science from UCSB. If you are interested in giving a presentation in this seminar series, or to make suggestions for speakers, please contact James Harland, the seminar co-ordinator. [an error occurred while processing this directive] |