Menu

Curriculum Vitae Download PDF file

Research Interests

My research interests are twofold: On one side I am interested in swarm intelligence, which is an artificial intelligence discipline based on the inspiration taken, for example, from the collective behaviour of social insects, flocks of birds, and fish schools. On the other side I am also interested in the hybridization of metaheuristics with more classical artificial intelligence and operations research methods such as, for example, branch and bound techniques and dynamic programming.

I am making use of swarm intelligence concepts both for solving challenging combinatorial optimization problems and for problem solving in distributed enviroments such as adhoc and sensor networks. Well-known swarm intelligence algorithms for combinatorial optimization are ant colony optimization (ACO) and particle swarm optimization (PSO). In distributed environments I have lately made use of the self-organization of natural ant colonies for obtaining synchronized sleep-wake periods, and the self-desynchronization of the calling periods of Japanese tree frogs. Concerning the second research line, I am currently working mainly on two different types of hybridization. First, the hybridization of metaheuristics based on the construction of solutions with concepts from branch and bound. Second, I am working on the development of efficient large neighborhood search algorithms.

Concerning applications, my work has a strong interdisciplinary flavour. In fact, optimization and control tasks arise in many important application areas such as telecommunications, bio-informatics, neuroscience, and robotics. A representative example of recent interdisciplinary work concerns the colboration with the Computer Vision Lab of the EPFL (Lausanne, Switzerland) on the automated reconstruction of dentritic and axonal trees. Concerning the bio-informatics (respectively, bio-medical) field, some of my work has focused on DNA sequencing, the training of neural networks for medical pattern classification, and on the founder sequence reconstruction problem.

Important lines of my work are shortly summarized below.

Hybrid Metaheuristics

Concerning the hybridization of metaheuristics with other techniques for optimization I am mainly working on two different types of hybrids. The first one is known as Beam-ACO. This is an algorithm that results from the combination of ant colony optimization with beam search, a branch and bound derivative. The second hybrid is known as large neighborhood search, which is a special type of local search that uses a complete method for exploring large-scale neighborhoods. The following two papers are recent examples for this line of research.

• A. Roli and S. Benedettini and T. Stuetzle and C. Blum. Large Neighbourhood Search Algorithms for the Founder Sequence Reconstruction Problem. Computers & Operations Research, 39(2):213--224, 2012.

• M. Lopez and C. Blum. Beam-ACO for the travelling salesman problem with time windows. Computers & Operations Research, 37(9):1570--1583, 2010.

Swarm Intelligence

Some of my recent work in swarm intelligence makes use of the self-synchronization observed in natural ant colonies for obtaining a mechanism for self-organized duty-cycling in sensor networks. Another example concerns the use of the calling behaviour of Japanese tree frogs for graph coloring.

• H. Hernandez and C. Blum. Foundations of ANTCYCLE: Self-Synchronized Duty-Cycling in Mobile Sensor Networks. The Computer Journal, 2011. In press.

• H. Hernandez and C. Blum. Distributed Graph Coloring: An Approach Based on the Calling Behavior of Japanese Tree Frogs. Submitted to Artificial Intelligence, 2011.

Applications

Apart from classical operations research applications, I have recently focused on interesting applications in wireless sensor networks and from the bioinformatics field. An interdisciplinary application from the Neuroscience field is shortly described in the following.

Tree-like structures, such as dendritic, vascular, or bronchial networks, are pervasive in biological systems. Despite of many years of work towards automated delineation techniques, the existing techniques remain fragile and error-prone. In this work, we use 3D optical micrographs of neurons and 2D retinal fundus images to demonstrate the importance of taking global tree structure and geometry into account to improve topological accuracy of the delineations.

The approach that we propose is based on ant colony optimization. It builds a set of candidate trees over many different subsets of points likely to belong to the optimal delineation and then chooses the best one according to a global objective function that combines image evidence with geometric priors.

More information can be otained at http://cvlab.epfl.ch/research/medical/lm/.

• E. Turetgen and G. Gonzalez and C. Blum and P. Fua. Automated Reconstruction of Dendritic and Axonal Trees by Global Optimization with Geometric Priors. Neuroinformatics, 9(2-3):279--302, 2011.

• E. Turetgen and C. Blum and G. Gonzalez and P. Fua. Reconstructing geometrically consistent tree structures from noisy images. In: Proceedings of MICCAI 2010 -- 13th International Conference on Medical Image Computing and Computer-Assisted Intervention., Lecture Notes in Computer Science, Springer, Berlin, Germany, 2010.

Survey Papers

• C. Blum and J. Puchinger and G. Raidl and A. Roli. Hybrid Metaheuristics in Combinatorial Optimization: A Survey. Applied Soft Computing, 11(6):4135--4151, 2011.

• C. Blum. Ant colony optimization: Introduction and recent trends. Physics of Life Reviews, 2(4):353-373, 2005.

• M. Dorigo and C. Blum. Ant colony optimization theory: A survey. Theoretical Computer Science, 344(2-3):243-278, 2005.

• C. Blum and A. Roli. Metaheuristics in combinatorial optimization: Overview and conceptual comparison. ACM Computing Surveys, 35(3):268-308, 2003.