JST

Tugas	Tgl Tugas	Tgl Selesai
Tugas 3	25 Maret 2010	5 Mei 2010

Nama	NPM	TUGAS
Dwi Ferdiana	0415031047	Mencari bahan, memasukan ke blog
Anisa Ulya D	0855031005	Mencari bahan, memasukan ke blog
Yuly Indriani	0815031097	memasukkan ke blog
Novia Utami P	0815031077	memasukan ke blog

% Prediksi Kuantitas Ekspor Daun Tembakau Indonesia Tahun 1996-2001

% Oleh : kelompok 5-A
% Anggota :
% Dwi Ferdiana
% Annisa Ulya Darajat
% Novia Utami Putri
% Yuli Indriani
%--------------------------------------------------

%=====INISIALISASI AWAL MULTILAYER PERCEPTRON=====
input = 5; %/ Jumlah input layer
hidden = 1; %/ Jumlah hidden layer
output = 1; %/ Jumlah output layer
lr = 0.5; %/ Learning Rate
miu = 1;%/Nilai miu
t1 = 0,33;% Target 1 yang diharapkan

% INPUT
x1=0,15;
x2=0,22;
x3=0,66;
x4=0,68;
x5=0,45;

% BOBOT INPUT KE HIDDEN
w11=0.1;
w12=0.2;
w13=0.3;
w14=0.1;
w15=0.2;
w21=0.3;
w22=0.1;
w23=0.2;
w24=0.3;
w25=0,1;

% BOBOT HIDDEN LAYER KE OUTPUT
v11=0.4;
v12=0.5;


% ITERASI --------
for iterasi= 1:2;

% =====FORWARD=====

% KELUARAN HIDDEN LAYER
h1 =(x1*w11)+(x2*w12)+(x3*w13)+(x4*w14)+(x5*w15);
h2 =(x1*w21)+(x2*w22)+(x3*w23)+(x4*w24)+(x5*w25);

H1 =(1/(1+exp(-miu*h1)));
H2 =(1/(1+exp(-miu*h2)));


% OUTPUT HIDDEN LAYER
y1 =(H1*v11)+(H2*v12)
Y1 =(1/(1+exp(-miu*y1)))


% =====BACKPROPAGATION=====

% EROR
e1=t1-Y1;


% UPDATE BOBOT OUTPUT KE HIDDEN DENGAN e1
dv11 = lr*e1*Y1*(1-Y1)*H1;
dv12 = lr*e1*Y1*(1-Y1)*H2;


v11 = v11 + dv11;
v12 = v12 + dv12;


% UPDATE BOBOT HIDDEN KE INPUT
dW11 = lr*((e1*Y1*(1-Y1)*v11))*H1*(1-H1)*x1
dW12 = lr*((e1*Y1*(1-Y1)*v11))*H1*(1-H1)*x2
dW13 = lr*((e1*Y1*(1-Y1)*v11))*H1*(1-H1)*x3
dW14 = lr*((e1*Y1*(1-Y1)*v11))*H1*(1-H1)*x4
dW15 = lr*((e1*Y1*(1-Y1)*v11))*H1*(1-H1)*x5
dW21 = lr*((e1*Y1*(1-Y1)*v12))*H2*(1-H2)*x1
dW22 = lr*((e1*Y1*(1-Y1)*v12))*H2*(1-H2)*x2
dW23 = lr*((e1*Y1*(1-Y1)*v12))*H2*(1-H2)*x3
dW24 = lr*((e1*Y1*(1-Y1)*v12))*H2*(1-H2)*x4
dW25 = lr*((e1*Y1*(1-Y1)*v12))*H2*(1-H2)*x5

end;


Hasil:

t1 =

0


x1 =

0


x2 =

0


x3 =

0


x4 =

0


x5 =

0


w25 =

0


y1 =

0.4500


Y1 =

0.6106


dW11 =

0


dW12 =

0


dW13 =

0


dW14 =

0


dW15 =

0


dW21 =

0


dW22 =

0


dW23 =

0


dW24 =

0


dW25 =

0


y1 =

0.4137


Y1 =

0.6020


dW11 =

0


dW12 =

0


dW13 =

0


dW14 =

0


dW15 =

0


dW21 =

0


dW22 =

0


dW23 =

0


dW24 =

0


dW25 =

0


Kesimpulan:
1.Jaringan syaraf tiruan akan belajar hingga nilai output yang diperoleh akan sama atau mendekati nilai target.
2. semakin banyak iterasi hasil akan semakin baik (nilai output mendekati nilai target).
3. Semakin banyak jumlah hidden layer akan semakin lama proses pembelajarannya akan tetapi hasil yang didapat akan semakin baik.