Personale docente

Stefano Cagnoni graduated in Electronic Engineering at the University of Florence in 1988. He has been a Ph.D. student and a post-doc until 1997, working in the Bioengineering Lab of the Department of Electronic Engineering. He received the Ph.D. degree in Bioengineering in 1993 from the same institution.
In 1994, he was a visiting scientist at the Whitaker College Biomedical Imaging and Computation Laboratory at the Massachusetts Institute of Technology. He held a post-doc position at the University of Florence in 1995 and 1996. Since 1997 he has been with the Department of Computer Engineering of the University of Parma, where he has been Associate Professor since 2004.
As concerns basic research, his main research interests are in soft computing, with particular regard to evolutionary computation and neural nets.
As concerns applied research, his main topics of interest are applying the above-mentioned techniques to problems in pattern recognition and computer vision.
As concerns neural nets, his research activity has concerned the development of modular neural net systems for computer vision applications and the sub-symbolic processing of non-numerical data.
In the field of evolutionary computation, besides actively participating in EvoNET, the EU-funded network of excellence in evolutionary computation (from 1997 to 2005, when the project was officially closed), he has developed applications of genetic algorithms to the design of filters for the analysis of one-dimensional signals and contour-based image segmentation. He has used Genetic Programming (GP) to create false-color images to synthesize the information in sequences of grey-level ones, using an interactive evolutionary method of which the user is an active component. GP has also been used to define the optimum play strategy for a robot goalkeeper, which successfully participated in RoboCup99. Using an original approach based on cellular programming, a license plate character recognition system has also been developed. Efficient GP algorithms have also been studied, particularly good at obtaining high-speed binary classification programs in a relatively short time. Using such techniques, he could obtain low-resolution character classifiers that showed a ten-fold increase in performance despite keeping almost the same accuracy as a reference neural classifier.
He has also studied the coevolution of heterogeneous systems, aimed, as a long-term goal, to define co-evolutionary algorithms based on the definition of ecosystems in which different populations evolve cooperatively and self-organize optimizes a specific part of the problem at hand.
His research has been focused on the theory and applications of Particle Swarm Optimization (PSO). In particular, as regards applications, he has studied computer vision algorithms in which PSO is used to detect and track objects/people and define a highly-efficient massively parallel implementation of PSO on graphics hardware.
More recently, he has applied Evolutionary Computation techniques implemented on GPUs to analyze complex systems, making it possible to compute the so-called Relevant Indices in a reasonable time using such techniques. The analysis methods based on the Relevant Indices have been used for feature construction in pattern recognition applications.
As regards computer vision, research has been dedicated mostly to low-level techniques and, in particular, to contour-based and region-based image segmentation, aimed, in a first phase, at extracting three-dimensional information from sequences of tomographic images, and, after that, at locating the free space in applications of autonomous navigation to robotics. Massively-parallel computer architectures based on cellular automata have been used to implement algorithms for perspective-effect removal and computation of the optical flow field. He has also designed a hybrid sensor prototype made up of a traditional camera associated with an omnidirectional one. Such a system can provide a 360-degree field of view at low resolution, along with a more limited field of view, but at a much higher resolution. These two situations correspond, roughly, to what happens in the human eye. The so-called peripheral vision (wide angle of view and low-resolution) coexists with the so-called foveal vision (high resolution in a neighborhood of the focus of attention). The contemporary presence of the two sensors makes it possible for stereo image processing to be performed. Such sensors have been applied to solve autonomous navigation problems in robotics and to surveillance tasks. More recently, he has studied metaheuristics to optimize contour-based segmentation based on elastic contour models.
Within the EU-funded MIBISOC project (see below), he has applied Evolutionary Computation to optimizing elastic contour methods for biomedical image segmentation.
More recently, he has applied Evolutionary Computation techniques implemented on GPUs to analyze complex systems, making it possible to compute the so-called Relevant Indices in a reasonable time using such techniques. The analysis methods based on the Relevant Indices have been used for feature construction in pattern recognition applications.
Professor Cagnoni's research activity has been carried out within projects financed by MIUR, CNR, ASI, and ENEA and part of contracts managed directly or within regional technology transfer projects between his group and firms in the province Parma or Emilia Romagna. He has co-managed a project funded by the Italian Railway Network Society (RFI) to develop an inspection system for train pantographs based on computer vision, which originated a patent whose rights have been acquired by a multinational corporation has further developed the project up to an industrial product level.
He has been awarded a grant from the EU, within the "Marie Sklodowska Curie" Actions, for a four-year research education project ("MIBISOC") in Medical Imaging using Bio-Inspired and Soft Computing.
He has organized workshops, continuing education courses, and edited special international journals; he frequently acts as a reviewer for international conferences and journals.
He has been a member of the Managing Board and secretary (2006-2007) of the Italian Association for Artificial Intelligence (AI*IA). He has been a member of the Advisory Board of Perada, the UE Coordination Action on Pervasive Adaptation. He is currently a member of the SPECIES (Society for the Promotion of Evolutionary Computation in Europe and Surroundings) advisory board.
From 1999 to 2018, he has been chairman of a yearly European event dedicated to evolutionary computation applications to image analysis and signal processing (now a track of EvoApplications, the European conference on the Applications of Evolutionary Computation, part of Evostar, a multi-conference on Evolutionary and Bio-inspired computation). In 2001 and 2002, he was General Chair of EvoWorkshops (presently EvoApplications)
He was the founder of GSICE, the Italian Workshop on Evolutionary Computation, in 2005 and chairman in 2006, before it became WIVACE (Italian - presently International - Workshop on Evolutionary Computation, Artificial Life and Complex Systems). He organized the Evonet Summer School on Evolutionary Computation in 2003 and SECEVITA, Summer School on Evolutionary Computation and Artificial Life, in 2007.
Chairman of the workshop on Evolutionary Computation, held at ECAI, the European Conference on Artificial Intelligence, in 2006. In 2012 and 2018, he organized WIVACE in Parma. In 2018, he was also the organizer of Evostar in Parma.
From 2005 to 2020, he has co-chaired MedGEC, a workshop on medical applications of evolutionary computation held concurrently with GECCO (Genetic and Evolutionary Computation Conference). Co-editor of the special issues of "EURASIP Journal of Applied Signal Processing" (July 2003), "Pattern Recognition Letters" (2006), "Evolutionary Computation" (2008), "Applied Soft Computing" (2020) dedicated to "Genetic and Evolutionary Computation for Image Analysis, Signal Processing, and Pattern Recognition." He organized the "EvoNET Summer School on Evolutionary Computation" in Parma in 2003. He has been Editor-in-chief of the "Journal of Artificial Evolution and Applications" (2007-2009).
He is a reviewer for many international journals and a member of the program committees of several conferences, among which: Evostar, Genetic and Evolutionary Computation Conference (GECCO), IEEE Conference on Evolutionary Computation (CEC), International Joint Conference on Artificial Intelligence (IJCAI).
He has been awarded the "Evostar 2009 Award", awarded by SPECIES to recognize the most outstanding contribution to Evolutionary Computation.
As of January 2021, his H-Index is 29 (Google) / 20 (Scopus).