DATE: Tuesday, July 5th, 2016
ROOM: WN-M639 (Science Building, VU Campus)
Semantic Genetic Programming Frameworks: Alignment in the Error Space and Equivalence classes
Semantics is one of the hottest topics in Genetic Programming. We explain the concepts of error vector and error space directly bound to semantics and, based on these concepts, we introduce the notions of optimally aligned individuals and optimally coplanar individuals. We show that, given optimally aligned individuals, it is possible to construct a globally optimal solution analytically (Error Space Alignment GP (ESAGP) algorithm). Using two complex real-life applications, we provide experimental evidence that exploiting alignments is useful and fast. During future work discussion we propose a framework based on Equivalence Classes that include ESAGP, Linear Scaling GP and other simple variants in a broader family of GP algorithm, and we show some preliminary results.
Stefano Ruberto is PhD student at the Gran Sasso Science Institute (L’Aquila,Italy), and he is working under the supervision of Leonardo Vanneschi (NOVA IMS, Universidade Nova de Lisboa, Portugal) and Ivano Malavolta (Vrije Universiteit Amsterdam).
You can find the slides of his talk here.