I am trying to solve a problem with 3 inputs and 1 output.
Data loaded using pandas .csv file.
train_test_split of the data done.
used standard scaler and minmax scaler.
I have made modifications in the 0318_Feed Forward Network as per my need of 3 inputs and 1 output with 1 hidden layer in between of 3 neurons. It ran w/o any problem.
class FirstFFNetwork:
def __init__(self):
self.w1 = np.random.randn()
self.w2 = np.random.randn()
self.w3 = np.random.randn()
self.w4 = np.random.randn()
self.w5 = np.random.randn()
self.w6 = np.random.randn()
self.w7 = np.random.randn()
self.w8 = np.random.randn()
self.w9 = np.random.randn()
self.w10 = np.random.randn()
self.w11 = np.random.randn()
self.w12 = np.random.randn()
self.b1 = 0
self.b2 = 0
self.b3 = 0
self.b4 = 0
self.b4 = 0
def sigmoid(self, x):
return 1.0/(1.0 + np.exp(-x))
def forward_pass(self, x):
self.x1, self.x2, self.x3 = x
self.a1 = self.w1*self.x1 + self.w2*self.x2 + self.w3*self.x3 + self.b1
self.h1 = self.sigmoid(self.a1)
self.a2 = self.w4*self.x1 + self.w5*self.x2 + self.w6*self.x3 + self.b2
self.h2 = self.sigmoid(self.a2)
self.a3 = self.w7*self.x1 + self.w8*self.x2 + self.w9*self.x3 + self.b3
self.h3 = self.sigmoid(self.a3)
self.a4 = self.w10*self.h1 + self.w11*self.h2 + self.w12*self.h3 + self.b4
self.h4 = self.sigmoid(self.a3)
return self.h4
def grad(self, x, y):
self.forward_pass(x)
self.dw10 = (self.h4-y) * self.h4*(1-self.h4) * self.h1
self.dw11 = (self.h4-y) * self.h4*(1-self.h4) * self.h2
self.dw12 = (self.h4-y) * self.h4*(1-self.h4) * self.h3
self.db4 = (self.h4-y) * self.h4*(1-self.h4)
self.dw1 = (self.h4-y) * self.h4*(1-self.h4) * self.w10 * self.h1*(1-self.h1) * self.x1
self.dw2 = (self.h4-y) * self.h4*(1-self.h4) * self.w10 * self.h1*(1-self.h1) * self.x2
self.dw3 = (self.h4-y) * self.h4*(1-self.h4) * self.w10 * self.h1*(1-self.h1) * self.x3
self.db1 = (self.h4-y) * self.h4*(1-self.h3) * self.w10 * self.h1*(1-self.h1)
self.dw4 = (self.h4-y) * self.h4*(1-self.h4) * self.w11 * self.h2*(1-self.h2) * self.x1
self.dw5 = (self.h4-y) * self.h4*(1-self.h4) * self.w11 * self.h2*(1-self.h2) * self.x2
self.dw6 = (self.h4-y) * self.h4*(1-self.h4) * self.w11 * self.h2*(1-self.h2) * self.x3
self.db2 = (self.h4-y) * self.h4*(1-self.h4) * self.w11 * self.h2*(1-self.h2)
self.dw7 = (self.h4-y) * self.h4*(1-self.h4) * self.w12 * self.h3*(1-self.h3) * self.x1
self.dw8 = (self.h4-y) * self.h4*(1-self.h4) * self.w12 * self.h3*(1-self.h3) * self.x2
self.dw9 = (self.h4-y) * self.h4*(1-self.h4) * self.w12 * self.h3*(1-self.h3) * self.x3
self.db3 = (self.h4-y) * self.h4*(1-self.h3) * self.w12 * self.h3*(1-self.h3)
def fit(self, X, Y, epochs=1, learning_rate=1, initialise=True, display_loss=False):
# initialise w, b
if initialise:
self.w1 = np.random.randn()
self.w2 = np.random.randn()
self.w3 = np.random.randn()
self.w4 = np.random.randn()
self.w5 = np.random.randn()
self.w6 = np.random.randn()
self.w7 = np.random.randn()
self.w8 = np.random.randn()
self.w9 = np.random.randn()
self.w10 = np.random.randn()
self.w11 = np.random.randn()
self.w12 = np.random.randn()
self.b1 = 0
self.b2 = 0
self.b3 = 0
self.b4 = 0
if display_loss:
loss = {}
for i in tqdm_notebook(range(epochs), total=epochs, unit="epoch"):
dw1, dw2, dw3, dw4, dw5, dw6, db1, db2, db3 = [0]*9
for x, y in zip(X, Y):
self.grad(x, y)
dw1 += self.dw1
dw2 += self.dw2
dw3 += self.dw3
dw4 += self.dw4
dw5 += self.dw5
dw6 += self.dw6
dw7 += self.dw7
dw8 += self.dw8
dw9 += self.dw9
dw10 += self.dw10
dw11 += self.dw11
dw12 += self.dw12
db1 += self.db1
db2 += self.db2
db3 += self.db3
db4 += self.db4
m = X.shape[1]
self.w1 -= learning_rate * dw1 / m
self.w2 -= learning_rate * dw2 / m
self.w3 -= learning_rate * dw3 / m
self.w4 -= learning_rate * dw4 / m
self.w5 -= learning_rate * dw5 / m
self.w6 -= learning_rate * dw6 / m
self.w7 -= learning_rate * dw7 / m
self.w8 -= learning_rate * dw8 / m
self.w9 -= learning_rate * dw9 / m
self.w10 -= learning_rate * dw10 / m
self.w11 -= learning_rate * dw11 / m
self.w12 -= learning_rate * dw12 / m
self.b1 -= learning_rate * db1 / m
self.b2 -= learning_rate * db2 / m
self.b3 -= learning_rate * db3 / m
self.b4 -= learning_rate * db4 / m
if display_loss:
Y_pred = self.predict(X)
loss[i] = mean_squared_error(Y_pred, Y)
if display_loss:
plt.plot(list(loss.values()))
plt.xlabel('Epochs')
plt.ylabel('Mean Squared Error')
plt.show()
def predict(self, X):
Y_pred = []
for x in X:
y_pred = self.forward_pass(x)
Y_pred.append(y_pred)
return np.array(Y_pred)
After running this code
ffn = FirstFFNetwork()
ffn.fit(X_train, Y_train, epochs=2000, learning_rate=.01, display_loss=True)
This is showing an error
23
24 def forward_pass(self, x):
---> 25 self.x1, self.x2, self.x3 = x
26 self.a1 = self.w1*self.x1 + self.w2*self.x2 + self.w3*self.x3 + self.b1
27 self.h1 = self.sigmoid(self.a1)
ValueError: too many values to unpack (expected 3)
