BPANNA Code
Status: Beta
Brought to you by:
gnusci
| File | Date | Author | Commit |
|---|---|---|---|
| bin | 2011-03-14 | gnusci | [r1] initial import |
| flws | 2011-11-30 | gnusci | [r36] WIN32 port |
| gui | 2013-07-09 | gnusci | [r42] using fltk-config for portability |
| include | 2012-05-21 | gnusci | [r41] minor bug in the Makefile |
| obj | 2011-03-14 | gnusci | [r1] initial import |
| prg | 2011-11-27 | gnusci | [r32] tmatrix bugs fixed |
| src | 2011-11-28 | gnusci | [r34] added activation function selection |
| History | 2011-11-30 | gnusci | [r39] Win32 port working |
| Makefile | 2011-03-14 | gnusci | [r3] minor modification of head files |
| README | 2011-03-14 | gnusci | [r1] initial import |
| TODO | 2011-11-25 | gnusci | [r11] minor modifications |
| license | 2011-03-14 | gnusci | [r3] minor modification of head files |
| makeinclude | 2013-07-09 | gnusci | [r42] using fltk-config for portability |
| version | 2011-11-24 | gnusci | [r8] added TODO |
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Directories
include: head files
fws: FLUID GUI development workspace
prg: test source codes programs
src: source files
gui: FLTK GUI interface (automatically generated by FLUID)
obj: temporal object files
bin: executable example programs
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Compilation instructions
Copy the compress file wherever you want to compile and untar it
> $ tar -xvzf fl_anna.tar.gz
> $ cd fl_anna
> $ make
you will find a test programs in the "bin" directory
just type
> ./bpanna
for no GUI interface or,
> ./fl_bpann
for a FLTK GUI application
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
DOCUMENTATION:
The basics steps are:
read_data_file(file); load a data file
set_output(no); Set number of outputs (no)
set_input(ni); Set number of inputs (ni)
set_layers(nl); Set number of layers (nl)
create(); create the neural network framework
set_neurons_in_layer(l,n); Set number (n) of neurons of the layer (l)
select_training_algorithm(b); true = Levenverg-Maquardt (LM),
false = Conjugate gradient (CG)
initialize(); initialize the neural network weights
set_alpha(alpha); this is the CG learning rate
set_mu(mu); this is the LM perturbation factor
training(); one step training
evalue(); evalue the net and give the result in a vector
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
The Artificial Neural Network Architecture (ANNA) is a
Back propagation neural network class easy to use with
the FLTK library. So you will need to have FLTK-1.3 installed.
You can get FLTK from www.fltk.org
The ANNA class library is compose by the following files:
anna_bpn_net.h anna_bpn_layer.h anna_bpn_neuron.h
anna_bpn_net.cxx anna_bpn_layer.cxx anna_bpn_neuron.cxx
The distribution include a demo which should work on Linux systems.
So enjoy it!
Edmanuel Torres
eetorres@gmail.com
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++