Nature, a rich source of inspiration

Dr. Christian Blum from the ALBCOM research group received recently the IEEE Transactions on Evolutionary Computation Outstanding Paper Award for the article " Search bias in ant colony optimization: On the role of competition-balanced systems" . This important prize will bring him recognition, a small economical reward and visibility. Today we are happy to spend some time asking him questions about his research and the prize.

Christian Blum has just been working at the Software Department for the last four years so the first question should be...
I was born in Germany. However, I have spent the last nine years in several different parts of Europe. I worked at the Imperial Cancer Research Fund (ICFR) in London. I pursued my PhD in Brussels and now I am here, in Barcelona. I am a "Ramon y Cajal" research fellow at the ALBCOM research group of the Universitat Politècnica de Catalunya (UPC).
I received my Diploma (Master of Science) in Mathematics in 1998 from the Kaiserlautern University in Germany. Later, I received my doctoral degree in Applied Science in 2004 from the Université Libre de Bruxelles in Belgium. My PhD Thesis was about Ant Colony Optimization.

Nowadays, I work mainly on metaheuristic methods to solve combinatorial optimization problems arising, for example, in telecommunication networks and adhoc networks.

Computational ants

The IEEE Transactions on Evolutionary Computation

Mathematics, computers and ants? I consider myself an applied mathematician. You could say that there is a huge overlap between mathematics and computing. In my opinion, mathematics and computing are both a tool and a language to problem solving What is very appealing in mixing mathematics, computers and ants is that you are inspired by nature and you are bringing all the knowledge you get from it into technology. I can say that I develop algorithms based on swarm intelligence mechanisms. Ant colony optimization (ACO) was introduced as a technique for combinatorial optimization in the early 1990s. The inspiring source of ACO is the foraging behaviour of real ant colonies. At the core of this behaviour is the indirect communication between ants by means of chemical pheromone trails, which enables them to find short paths between their nest and food sources. This characteristic of real ant colonies is exploited in ACO algorithms in order to solve, for example, discrete and continuous optimization problems, and to manage tasks such as routing and load balancing in telecommunication networks and, more recently, in mobile adhoc networks such as sensor networks. What is your research about? My main research interests are in swarm intelligence methods for optimization and in the hybridization of metaheuristics with more classical techniques for optimization.  I am mainly focus on a technique called ant colony optimization (ACO), whose inspiring source is the shortest path finding ability of real ant colonies. In terms of applications, my interests are in optimization problems from the fields of telecommunications and bio-informatics. Why do you use hybrid metaheuristic techniques? Human beings know more and more. We organise ourselves in a more complex way, we develop new, bigger and better ideas and applications. All these facts make problems become bigger. Large numbers of choices are possible and traditional ways of calculation and organisation are not enough anymore. Traditional techniques ("or exact techniques") are precise. However, often this results in an enormous calculation time. On the other hand, metaheuristic techniques can give you an approximated quick solution. A combination of both techniques reduces time and increases accuracy. What about logics? It is true that some researchers use logics for scheduling or optimization issues in general. However, their initial situations are slightly different from ours. They usually work with many restrictive conditions and they have to find a solution that satisfies all constraints. We would not have a proper answer for those situations due to the approximative style of our solutions. On the other hand, in a similar fashion as we do when  we combine exact techniques with metaheuristic techniques, there are some researchers who have started to combine metaheuristics with logics. Am I right if I see a relationship between your research and Artificial Intelligence? Sure. All these ways to approach problems have common structures. I also make use of some classical Artificial Intelligence and Operations Research methods such as, for example, Branch & Bound techniques or Dynamic Programming.

What is the IEEE Transaction on Evolutionary Computation? The IEEE Transactions on Evolutionary Computation (TEC) is a monthly journal published by the Computational Intelligence Society of the IEEE Computer Society. It contains peer-reviewed articles and other contributions in the area of evolutionary computation and natural computation. It is intended for researchers, developers, educators and technical managers in the computer field. Why is so important to publish in it? Because it is widely considered to be the leading journal in the area. The magazine ranked among the most-cited journals (number 2 in Computer Science, Theory & Methods and number 3 in Artificial Intelligence), according to the 2007 annual Journal Citation Report, published by the Thomson Institute for Scientific Information. Being one of the best and more popular journals, it is quite hard to get an article published in it. Hard but possible and now a reality! Why do you think they liked your article? Optimization techniques based on swarm intelligence have become increasingly popular during the last decade. They are characterized by a decentralized way of working that mimics the behaviour of swarms of social insects, flocks of birds, or fish schools. The advantage of these approaches over traditional techniques is their robustness and flexibility. These properties make swarm intelligence a successful design paradigm for algorithms that deal with increasingly complex problems. Most of the time this procedure works very well. However there are a few situations where this technique does not work. In 2005 I wrote an article that dealt with those odd situations where you can not use this way of working. It was called Search bias in ant colony optimization: On the role of competition-balanced systems. In my opinion, it is equally important to have a powerful theory as it is to know under what conditions you could use this theory. That is probably the reason why I got now, two years later, the best paper award. What does this award represent? It means an economical prize of 1000$, the recognition of the science community  and visibility. With this new perspective, how do you see your  future in research? At the moment I can only just think about my immediate future. I would like to explore in broader terms the applications of swarm intelligence techniques, not only to optimization but also to other tasks, for example, to the management of large-scale sensor mothods. From this point of view it is fortunate that the ALBCOM reserach group has two European projects on sensors methods and, more generaly, on tiny artifacts.

ilapuente@lsi.upc.edu

(Back to the Newsletter)