The Fourth International Workshop on Computational Intelligence in Economics and Finance (CIEF 2005) will be held as a part of the Eighth Joint Conference on Information Sciences. This series of events originated four years ago in Atlantic City (CIEF 2000), and was followed by two consecutive workshops in Research Triangle Park, North Carolina, (CIEF 2002, CIEF 2003). The series started with a small group of papers -- 33 submitted and 26 accepted -- at CIEF 2000, and then grew to 85 papers (74 accepted) at CIEF 2003. While all papers were initially submitted as four-page extended abstracts, many participants were invited to revise these abstracts and submit them again as full papers. Some of them have finally been accepted and published in a special book volume or journal issue. Among them, CIEF 2000 was published as a special issue in the journal Information Sciences, and CIEF 2002 was published in the form of a book entitled "Computational Intelligence in Economics and Finance" (Springer, 2003). CIEF 2003 is now being prepared both as a book volume "Computational Economics: A Perspective from Computational Intelligence (IGI, http://www.idea-group.com)" and as a special issue of the journal Information Sciences.
The purpose behind CIEF is essentially to acknowledge the legacy to economics of Herbert Simon, who initiated the interdisciplinary research focus and broke down the conventional distinctions among economics, computer science and cognitive psychology. The last two played almost no role in the early days of Herbert Simon, but have now proved to be an indispensable part of economics, in particular when agent-based computational economics is now emerging as an integration of the originally disparate research on experimental economics, behavioral finance and economics with heterogeneous interacting agents. The increasingly large interdisciplinary framework really leads us to an even richer environment than in the days of Herbert Simon.
The idea of computational intelligence is basically to model the intelligent behavior observed from linguistic behavior, biology, insects (swarm intelligence), neural sciences, and immune systems, as it is well said that "natural does it all". This is different from the classical AI, which was mainly motivated by and built upon mathematical logic. This is also different from the conventional models of learning, which are mainly based upon probability and statistics. However, modeling intelligence observed from natural behavior often leads us to computationally-intensive models because the subjects that we are trying to model are by no means as simple as the classical dynamic systems. We anticipate that, if we can model this observed behavior successfully, we can then have a better chance of understanding the operation of the economic system as a complex adaptive system, which is already a research target of many well-known research institutes.
With this expectation in mind, CIEF 2000 started to attract papers on novel economic and financial applications of computational intelligence. We began with those basic tools that every standard CI textbook should cover, namely, fuzzy logic, artificial neural networks, and evolutionary computation. Nevertheless, the ever-increasing interdisciplinary network has made us quickly realize that it would be meaningless to draw a clear line like this in such a complex and dynamically-evolving scientific environment. So, CIEF is also adapting. Intelligent behavior (and learning behavior as a part of it) is still our focus, but is not necessarily restricted to computational intelligence, be it narrowly defined or broadly defined. Models that are motivated by computational theory, statistics, econometrics, physics, mathematics, and psychology are all welcome.
When we accept papers from larger potential areas, we find that there are many conference participants who are interested in CI, but have little idea of CI. In addition, CI has now become quite a large family. Even CI specialists are not supposed to be familiar with all kinds of CI tools; therefore, to enhance the communication between us, beginning in CIEF 2003, we included tutorials as part of the conference program so that participants could learn some basics before they could go deeper. This year, tutorial proposals are still encouraged. Due to budget constraints, we cannot guarantee anything at this moment, but we will still do our best to support some of the best tutorial proposals. As in the past, one tutorial session is scheduled to last one hour, and the speaker can decide how many sessions are needed. Those who are interested in submitting a tutorial proposal instead of a regular paper should write to Prof. Chueh-Yung Tsao at email@example.com.
The second focus of CIEF is the application of CI to modeling the autonomous agents in agent-based computational models of economics and finance. So, agent-based computational economics and finance has become another broadly-defined interest of this conference. Since agent-based modeling has now also become a very popular tool in the management sciences, we also solicit papers on agent-based models of management. Even more, given the growing field of agent-based computational social sciences, CIEF 2005 is also soliciting papers on agent-based computational social sciences. While CI is a key weapon for agent engineering, other approaches, such as the formal approach or the analytical approach to agent-based modeling, are also acceptable. In fact, the most striking example seen in the last few CIEFs is the approach coming from Econphysics.
CIEF believes that the trend whereby agent-based modeling becomes integrated with various areas will continue. The artificial financial market has already become an area that is being intensively addressed. That is the reason why we have Prof. Blake LeBaron as the keynote speaker this year. Other areas, such as rational-expectations macroeconomics, experimental economics, and game theory, are also receiving increased attention. However, there are some other areas where little agent-based modeling has been done, but where there is much potential. Examples include network economics, industrial organization, behavioral finance, growth theory, human capital accumulation, evolutionary economics (preference and technology evolution), etc. Therefore, papers that do not explicitly fall in either the track of CI or agent-based modeling, but may highlight promising new applications of CI to economics, finance, or, generally, the social sciences, are also welcome.