CHAPTER7\ENVIClassicExtensions\
ffn3kal__define.pro
Object class for implementation of a three-layer, feed-forward neural network for classification of multi-spectral images with AdaBoost.M1 or .M2 boosting
Implements Kalman filter training.
Refs: Shaw and Palmieri, Proc. Int. Joint Conf. on Neural Networks, San Diego (1990) I(3), 41-46
Freund and Shapire, J. Comp. and Syst. Sci. 55, 1997, 119-139 Schwenk and Bengio, Neural Computation 12(8) 2000, 1869-1887
ffn = Obj_New("FFN3KAL",Xs,Ls,L1,L2,D=D,epochs=epochs,M2=M2)
Xs: array of observation column vectors observations are in [0,1]
Ls: array of class label column vectors of form (0,0,1,0,0,...0)^T
L1,L2: numbers of hidden neurons
D: adaboost sample weight distribution
epochs: numberof training epochs
M2: if 0, then AdaBoost.M1 (default), else M2
Class description for FFN3KAL
Inheritance
Properties
Properties in FFN3KAL
- epochs init
- D init
Fields
Fields in FFN3
- KK 0L
- W2H ptr_new()
- NN 0L
- NP 0L
- WO ptr_new()
- LS ptr_new()
- N1 ptr_new()
- N2 ptr_new()
- GS ptr_new()
- LL1 0L
- LL2 0L
- W1H ptr_new()
Fields in FFN3KAL
- S2H ptr_new()
- SD ptr_new()
- V ptr_new()
- SO ptr_new()
- S1H ptr_new()
- ITERATIONS 0L
- COST_ARRAY ptr_new()
- D ptr_new()
Author information
- Author
Mort Canty (2009) Juelich Research Center m.canty@fz-juelich.de
Other file information
- Uses:
PROGRESSBAR
Routines
result = FFN3KAL::Init(Gs, Ls, L1, L2 [, epochs=epochs], D=D)
Object class for implementation of a three-layer, feed-forward neural network classifier.
FFN3KAL::Cleanup
FFN3KAL::train, verbose=verbose, cancel=cancel
result = FFN3KAL::cost()
FFN3KAL__Define
Routine details
top FFN3KAL::Init
result = FFN3KAL::Init(Gs, Ls, L1, L2 [, epochs=epochs], D=D)
Object class for implementation of a three-layer, feed-forward neural network classifier. Implements Kalman filter training:
Shah, S. and Palmieri, F. (1990). Meka — A fast,
local algorithm for training feed forward neural
networks. Proceedings of the International Joint
Conference on Neural Networks, San Diego, I(3),
41–46.
Parameters
- Gs in required
array of observation column vectors
- Ls in required
array of class label column vectors of form (0,0,1,0,0,...0)^T
- L1 in required
number of hidden neurons
- L2
Keywords
- epochs in optional
training epochs per network in emsemble
- D in required
training sample distribution
Examples
ffn = Obj_New("FFN3KAL",Gs,Ls,L1,L2)
Author information
- Author:
Mort Canty (2009)
Other attributes
- Uses:
COYOTE
top FFN3KAL::Cleanup
FFN3KAL::Cleanup
top FFN3KAL::cost
result = FFN3KAL::cost()
top FFN3KAL__Define
FFN3KAL__Define
File attributes
Modification date: | Tue Nov 18 16:00:05 2014 |
Lines: | 263 |
Docformat: | rst rst |