DocumentCode :
2708177
Title :
Using the symmetries of a multi-layered network to reduce the weight space
Author :
Jordan, Frédéric ; Clement, Guillaume
Author_Institution :
Inst. Nat. des Sci. Appl., Rennes, France
fYear :
1991
fDate :
8-14 Jul 1991
Firstpage :
391
Abstract :
The results are presented of a theoretical study of multilayered neural networks carried out using a general formalism describing such networks, forward pass, backpropagation and coherent transformations. A weight space reducing method using sign and permutation transformations was developed. After remarking that certain network modifications (notably any permutation of two units in the same layer) have no effect on the global transfer function, the authors formalize and generalize this observation. Then, they demonstrate that this result could be used to reduce the search for solutions to a restricted part of the weighted space. Finally, a learning algorithm inspired by simulated annealing has made it possible to test the method
Keywords :
learning systems; neural nets; simulated annealing; transfer functions; backpropagation; coherent transformations; forward pass; global transfer function; learning algorithm; multilayered neural networks; permutation transformations; search reduction; sign transformations; simulated annealing; symmetries; weight space reducing method; Jacobian matrices; Neurons; Simulated annealing; Testing; Transfer functions;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Neural Networks, 1991., IJCNN-91-Seattle International Joint Conference on
Conference_Location :
Seattle, WA
Print_ISBN :
0-7803-0164-1
Type :
conf
DOI :
10.1109/IJCNN.1991.155365
Filename :
155365
Link To Document :
بازگشت