Title :
Learning method based on minimization of knowledge representation cost
Author_Institution :
Lab. de Recherche en Inf., Univ. de Paris-Sud, Orsay, France
Abstract :
The author introduces a practical method based on optimization of numerical criteria to define an interpretation task and a graph clustering task to learn intermediate abstractions as generalizations of similar subgraph groups. The dynamic of the system is organized as an evolutionary process which is based on genetic algorithms. The set of abstractions used in the conceptual graph constitutes the population to evolve in order to optimize the performance of the interpretation task
Keywords :
genetic algorithms; graph theory; knowledge representation; learning systems; conceptual graph; evolutionary process; genetic algorithms; graph clustering task; intermediate abstractions; interpretation task; knowledge representation cost; machine learning; minimization; numerical criteria; optimization; Computer vision; Cost function; Data analysis; Genetic algorithms; Knowledge representation; Learning systems; Machine learning; Minimization methods; Optimization methods;
Conference_Titel :
Intelligent Control, 1991., Proceedings of the 1991 IEEE International Symposium on
Conference_Location :
Arlington, VA
Print_ISBN :
0-7803-0106-4
DOI :
10.1109/ISIC.1991.187387
