Title :
Local component analysis: a neural model for information retrieval
Author_Institution :
CNRS, Paris, France
Abstract :
A neural model is presented for implementing an associative mapping of highly multidimensional data, as encountered in information retrieval systems, on a structured set of half-axes. The nonsupervised learning rule uses a local reconstitution of the data by the network and an analytical expression of the gain coefficient. The author shows first that the model can be configured for a stochastic approximation of principal component analysis and correspondence analysis. Then he discusses its application, in its associative mapping configuration, to real-world documentary databases. Finally, an example of such an application is presented.<>
Keywords :
information retrieval; learning systems; neural nets; associative mapping; correspondence analysis; documentary databases; gain coefficient; information retrieval systems; local component analysis; neural model; nonsupervised learning rule; principal component analysis; stochastic approximation; structured set; Information retrieval; Learning systems; Neural networks;
Conference_Titel :
Neural Networks, 1989. IJCNN., International Joint Conference on
Conference_Location :
Washington, DC, USA
DOI :
10.1109/IJCNN.1989.118676
