Semantics and Fitness Landscapes in Genetic Programming

26 February 2013

Leonardo Vanneschi ISEGI/INESC-ID

Moraglio et al. have recently introduced new genetic operators for genetic programming, called geometric semantic operators. It is possible to prove that these operators induce a unimodal fitness landscape for all the problems consisting in matching input data with known target outputs (like regression and classification). This feature facilitates genetic programming evolvability, which makes these operators extremely promising. Nevertheless, Moraglio et al. leave one big open problem: these operators, by construction, always produce offspring that are larger than their parents, causing an exponential growth in the size of the individuals, which actually renders them useless in practice.

In this seminar, a general introduction to optimization, to fitness landscapes and to evolutionary computation, with particular focus on genetic programming, is offered. After that, geometric semantic operators are presented and the fact that they induce a unimodal fitness landscape on every possible instance of regression and classification is informally shown. Finally, after discussing the limitation of geometric semantic operators, a new efficient implementation of them is presented.

This new implementation allows us, for the first time, to use these operators on complex real-life applications, like the two complex problems in pharmacokinetics that are discussed in the seminar. The presented experiments confirm the excellent evolvability of geometric semantic operators, demonstrated by the good results reported on training data. Furthermore, a surprisingly good generalization ability was achieved, testified by the excellent results obtained on test data.

This fact can be explained considering some particular properties of geometric semantic operators, which makes them even more appealing than before.



Leonardo Vanneschi obtained his Master Degree ("Laurea") in Computer Science at the University of Pisa (Italy) in 1996 ("summa cum Laude"), and his PhD in Computer Science at the University of Lausanne (Switzerland) in 2004 (PhD thesis honoured with the excellence award of the Science Faculty of the University of Lausanne). He is Assistant Professor at ISEGI, Universidade Nova de Lisboa (Lisbon, Portugal), since September 2011. His main Research interests are Machine Learning, the study of Complex Systems, and in particular Evolutionary Computation and Genetic Programming. During his career, he has established various collaborations, both national and internationals, participating in about 10 financed research projects. He is currently member of the editorial board of two of the most renown scientific journals in the field of Evolutionary Computation. He is member of the steering committee and program committee of various international conferences. He is and he has been the teacher of various courses in Computer Science at various different University levels. Until nowadays, he has published 120 contributions, among which 22 on scientific journals with impact factor and 8 have been honoured with international awards.