Luciano Costa's Group - home

Presentation

Luciano Costa's research group initiated in 1992, after his return from a PhD at King's College, University of London. Since then, the group has emphasized educational training and integration between multidisciplinary researches of basic and applied orientations. The works developed in the group are organized among three main issues: (a) pattern recognition and AI; (b) image processing and analysis; and (c) complex systems and networks. As a consequence of the generality of the concepts and methods in these areas, the works developed in the group acquired a pronounced inter and multidisciplinary nature, covering a wide diversity of fields such as arts, linguistics, neuroscience, economy, scientometry, systems biology, material sciences and agronomy.

While the group background is centered on the above mentioned three areas, the development of such a broad range of applications was only allowed through regular interactions with several researchers from other areas and institutions, including Harvard Medical School, Univ. California Univ. of Cambridge, Salk Institute, Univ. of London, Cold Spring Harbor, among others. In Brazil, interactions have involved IME-USP, ICMC-USP, Faculty of Medicine USP, EMBRAPA and Univ. of Uberlandia, among others.

The research results achieved in the group cover from new concepts and methods to diverse applications. Some research lines of particular success include the characterization, classification and simulation of morphological neuronal systems, with emphasis on the shape of neurons, their interconnections, and dynamics. Several methods have also been developed in the area of image and shape analysis, including multi scale skeletons, self-referred lacunarity, and methods for multi scale curvature estimation. In the field of complex systems and networks, we emphasize the analysis of several types of theoretical and real-world networks, including urban structures, Wikipedia, scientific citations, amongst others. Particular attention has been given to measurement and classification of complex networks, including the proposal of new measurements such as the accessibility and topological symmetry. Another subject of special interest in the group has been the topological characterization and modeling of complex networks. In patter recognition, we highlight works aimed at systematic comparison of classifiers using large and diverse types of synthetic data.

The impact and visibility of the works developed in the group are substantiated by about 300 journal articles and a book, which have motivated more than 10000 citations in Scholar and about 4000 citations in the Web of Science. A great deal of such expressive results is a consequence of the involvement of several students, through more than 50 graduate supervisions, whose quality has been acknowledged through the Prize Best PhD Thesis in physical sciences at USP in 2014, as well as by the hiring of trained researchers as lecturers at USP and other expressive institutions. Works developed in the group have appeared in journals such as Scientific American, Science, Nature Communications, Nature Neuroscience, Advances in Physics, ACM Surveys, Physical Review Letters, Neuron, eLife, and Journal of Neuroscience, to name but a few.

The areas of scientific dissemination and teaching are also actively supported by the group, including publications in magazines (e.g. Ciencia Hoje), several interviews, FAPESP magazine and Bulletin, Scientific Initiation Meetings (including awards) and the Undergraduate Congress of USP.

The group remains open for new collaborations and supervisions at the levels of scientific initiation, master, PhD, and post-doc. We observe that it is, in principle, possible to start MSc and PhD programs at any time of the year. Those interested can contact Luciano Costa through his email ldfcosta@gmail.com.

Inspirational