The purpose of this work is the development of a system able to recognize faces in different perspectives, introducing some kind of noise and having different expressions. The problem of faces recognition has been approached using a complex architecture based on a hierarchy of neural networks and on a particular self-referencing structure. The system, in fact, is structured as a tree in which nodes correspond to neural networks each one having different tasks and each leaf is a recognition module composed by some networks with different characteristics depending on the different preprocessing operators used. These networks are coordinated by a supervisor in a self-referencing structure; the supervisor, called Meta-Net, in training phase observes the behaviour of recognition nets while in test phase it decides, given an input image, which weights to assign to each network and modifies their output in order to obtain the final result. This architecture shows a high generalization capability and allows the recogni¬tion of images with different kind of noise.
Flocchini, P., Gardin, F., Mauri, G., Pensini, M., Stofella, P. (1991). Recognizing faces with a massively parallel system. In Proceedings PaCT91 - International Conference on Parallel Computing Technologies (pp.203-214). SGP : World Scientific.
Recognizing faces with a massively parallel system
MAURI, GIANCARLO;
1991
Abstract
The purpose of this work is the development of a system able to recognize faces in different perspectives, introducing some kind of noise and having different expressions. The problem of faces recognition has been approached using a complex architecture based on a hierarchy of neural networks and on a particular self-referencing structure. The system, in fact, is structured as a tree in which nodes correspond to neural networks each one having different tasks and each leaf is a recognition module composed by some networks with different characteristics depending on the different preprocessing operators used. These networks are coordinated by a supervisor in a self-referencing structure; the supervisor, called Meta-Net, in training phase observes the behaviour of recognition nets while in test phase it decides, given an input image, which weights to assign to each network and modifies their output in order to obtain the final result. This architecture shows a high generalization capability and allows the recogni¬tion of images with different kind of noise.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.