def loss(self, X, y=None, reg=1e-5):
print 'start computing loss and grad.............'
W1, b1 = self.params['W1'], self.params['b1']
W2, b2 = self.params['W2'], self.params['b2']
W3, b3 = self.params['W3'], self.params['b3']
# pass conv_param to the forward pass for the convolutional layer
filter_size = W1.shape[2]
conv_param = {'stride': 1, 'pad': (filter_size - 1) / 2}
# pass pool_param to the forward pass for the max-pooling layer
pool_param = {'pool_height': 2, 'pool_width': 2, 'stride': 2}
# compute the forward pass
print 'compute the forward pass......'
print 'compute the w1 conv_relu_pool_forward forward pass......'
a1, cache1 = layers.conv_relu_pool_forward(X, W1, b1, conv_param,
pool_param)
print 'compute the w2 affine_relu_forward forward pass......'
a2, cache2 = layers.affine_relu_forward(a1, W2, b2)
print 'compute the w3 affine_forward forward pass......'
scores, cache3 = layers.affine_forward(a2, W3, b3)
if y is None:
return scores
# compute the backward pass
print 'compute the backward pass......'
print 'compute the softmax_loss backward pass......'
data_loss, dscores = layers.softmax_loss(scores, y)
print 'compute the dw3 affine_backward backward pass......'
da2, dW3, db3 = layers.affine_backward(dscores, cache3)
print 'compute the dw2 affine_relu_backward backward pass......'
da1, dW2, db2 = layers.affine_relu_backward(da2, cache2)
print 'compute the dw1 conv_relu_pool_backward backward pass......'
dX, dW1, db1 = layers.conv_relu_pool_backward(da1, cache1)
# Add regularization
dW1 += self.reg * W1
dW2 += self.reg * W2
dW3 += self.reg * W3
reg_loss = 0.5 * self.reg * sum(np.sum(W * W) for W in [W1, W2, W3])
loss = data_loss + reg_loss
grads = {
'W1': dW1,
'b1': db1,
'W2': dW2,
'b2': db2,
'W3': dW3,
'b3': db3
}
print ' computing loss and grad end !!!!!!!!!!!!!!!!!'
print 'loss is :', loss
return loss, grads
def loss(self, X, y=None, reg=1e-5):
W1, b1 = self.params['W1'], self.params['b1']
W2, b2 = self.params['W2'], self.params['b2']
W3, b3 = self.params['W3'], self.params['b3']
# pass conv_param to the forward pass for the convolutional layer
filter_size = W1.shape[2]
conv_param = {'stride': 1, 'pad': (filter_size - 1) / 2}
# pass pool_param to the forward pass for the max-pooling layer
pool_param = {'pool_height': 2, 'pool_width': 2, 'stride': 2}
# compute the forward pass
a1, cache1 = layers.conv_relu_pool_forward(X, W1, b1, conv_param,
pool_param)
norm_out, norm_cache = layers.spatial_batchnorm_forward(
a1, 1, 0, bn_param={'mode': 'train'})
a2, cache2 = layers.affine_relu_forward(norm_out, W2, b2)
scores, cache3 = layers.affine_forward(a2, W3, b3)
if y is None:
return scores
# compute the backward pass
data_loss = NUS_loss_test.NUSDataTrain().loss(scores, y)
dscores = NUS_loss_test.NUSDataTrain().eval_numerical_gradient(
NUS_loss_test.NUSDataTrain().grad_loss,
scores) # layers.softmax_loss(scores, y)#改这里
da2, dW3, db3 = layers.affine_backward(dscores, cache3)
da1, dW2, db2 = layers.affine_relu_backward(da2, cache2)
dnorm_out, dgamma, dbeta = layers.spatial_batchnorm_backward(
da1, norm_cache)
dX, dW1, db1 = layers.conv_relu_pool_backward(dnorm_out, cache1)
# Add regularization
dW1 += self.reg * W1
dW2 += self.reg * W2
dW3 += self.reg * W3
reg_loss = 0.5 * self.reg * sum(np.sum(W * W) for W in [W1, W2, W3])
loss = data_loss + reg_loss
grads = {
'W1': dW1,
'b1': db1,
'W2': dW2,
'b2': db2,
'W3': dW3,
'b3': db3
}
return loss, grads
def loss(self,X,y=None):
"""
Compute loss and gradient for a minibatch of data.
Inputs:
- X: Array of input data of shape (N, d_1, ..., d_k)
- y: Array of labels, of shape (N,). y[i] gives the label for X[i].
Returns:
If y is None, then run a test-time forward pass of the model and return:
- scores: Array of shape (N, C) giving classification scores, where
scores[i, c] is the classification score for X[i] and class c.
If y is not None, then run a training-time forward and backward pass and
return a tuple of:
- loss: Scalar value giving the loss
- grads: Dictionary with the same keys as self.params, mapping parameter
names to gradients of the loss with respect to those parameters.
"""
scores = None
W1,b1 = self.params['W1'],self.params['b1']
W2,b2 = self.params['W2'],self.params['b2']
ar1_out,ar1_cache = affine_relu_forward(X,W1,b1)
ar2_out,ar2_cache = affine_forward(ar1_out,W2,b2)
scores = ar2_out
if y is None:
return scores
loss,grads = 0,{}
loss,dout = softmax_loss(scores,y)
loss = loss+0.5*self.reg*np.sum(W1*W1)+0.5*self.reg*np.sum(W2*W2)
dx2,dw2,db2 = affine_backward(dout,ar2_cache)
grads['W2'] = dw2 +self.reg*W2
grads['b2'] = db2
dx1,dw1,db1 = affine_relu_backward(dx2,ar1_cache)
grads['W1'] = dw1+self.reg*W1
grads['b1'] = db1
return loss,grads
def predict(self, X):
"""
Inputs:
- X: A numpy array of shape (N, D) giving N D-dimensional data points to
classify.
Returns:
- y_pred: A numpy array of shape (N,) giving predicted labels for each of
the elements of X. For all i, y_pred[i] = c means that X[i] is
predicted to have class c, where 0 <= c < C.
"""
y_pred = None
# h1 = layers.ReLU(np.dot(X, self.params['W1']) + self.params['b1'])
# scores = np.dot(h1, self.params['W2']) + self.params['b2']
# y_pred = np.argmax(scores, axis=1)
W1, b1 = self.params['W1'], self.params['b1']
W2, b2 = self.params['W2'], self.params['b2']
W3, b3 = self.params['W3'], self.params['b3']
# pass conv_param to the forward pass for the convolutional layer
filter_size = W1.shape[2]
conv_param = {'stride': 1, 'pad': (filter_size - 1) / 2}
# pass pool_param to the forward pass for the max-pooling layer
pool_param = {'pool_height': 2, 'pool_width': 2, 'stride': 2}
# compute the forward pass
a1, cache1 = layers.conv_relu_pool_forward(X, W1, b1, conv_param,
pool_param)
a2, cache2 = layers.affine_relu_forward(a1, W2, b2)
scores, cache3 = layers.affine_forward(a2, W3, b3)
y_pred = np.argmax(scores, axis=1)
return y_pred
def loss(self, X, y=None):
"""
Compute loss and gradient for the fully-connected net.
Input / output: Same as TwoLayerNet above.
"""
X = X.astype(self.dtype)
mode = 'test' if y is None else 'train'
# Set train/test mode for batchnorm params and dropout param since they
# behave differently during training and testing.
if self.dropout_param is not None:
self.dropout_param['mode'] = mode
if self.use_batchnorm:
for bn_param in self.bn_params:
bn_param['mode'] = mode
scores = None
############################################################################
# TODO: Implement the forward pass for the fully-connected net, computing #
# the class scores for X and storing them in the scores variable. #
# #
# When using dropout, you'll need to pass self.dropout_param to each #
# dropout forward pass. #
# #
# When using batch normalization, you'll need to pass self.bn_params[0] to #
# the forward pass for the first batch normalization layer, pass #
# self.bn_params[1] to the forward pass for the second batch normalization #
# layer, etc. #
############################################################################
layer_input = X
ar_cache = {}
dp_cache = {}
for lay in xrange(self.num_layers - 1):
if self.use_batchnorm:
layer_input, ar_cache[lay] = affine_bn_relu_forward(layer_input,
self.params['W%d' % (lay + 1)],
self.params['b%d' % (lay + 1)],
self.params['gamma%d' % (lay + 1)],
self.params['beta%d' % (lay + 1)],
self.bn_params[lay])
else:
layer_input, ar_cache[lay] = affine_relu_forward(layer_input, self.params['W%d' % (lay + 1)],
self.params['b%d' % (lay + 1)])
if self.use_dropout:
layer_input, dp_cache[lay] = dropout_forward(layer_input, self.dropout_param)
ar_out, ar_cache[self.num_layers] = affine_forward(layer_input, self.params['W%d' % (self.num_layers)],
self.params['b%d' % (self.num_layers)])
scores = ar_out
# pass
############################################################################
# END OF YOUR CODE #
############################################################################
# If test mode return early
if mode == 'test':
return scores
loss, grads = 0.0, {}
############################################################################
# TODO: Implement the backward pass for the fully-connected net. Store the #
# loss in the loss variable and gradients in the grads dictionary. Compute #
# data loss using softmax, and make sure that grads[k] holds the gradients #
# for self.params[k]. Don't forget to add L2 regularization! #
# #
# When using batch normalization, you don't need to regularize the scale #
# and shift parameters. #
# #
# NOTE: To ensure that your implementation matches ours and you pass the #
# automated tests, make sure that your L2 regularization includes a factor #
# of 0.5 to simplify the expression for the gradient. #
############################################################################
loss, dscores = softmax_loss(scores, y)
dhout = dscores
loss = loss + 0.5 * self.reg * np.sum(
self.params['W%d' % (self.num_layers)] * self.params['W%d' % (self.num_layers)])
dx, dw, db = affine_backward(dhout, ar_cache[self.num_layers])
grads['W%d' % (self.num_layers)] = dw + self.reg * self.params['W%d' % (self.num_layers)]
grads['b%d' % (self.num_layers)] = db
dhout = dx
for idx in xrange(self.num_layers - 1):
lay = self.num_layers - 1 - idx - 1
loss = loss + 0.5 * self.reg * np.sum(self.params['W%d' % (lay + 1)] * self.params['W%d' % (lay + 1)])
if self.use_dropout:
dhout = dropout_backward(dhout, dp_cache[lay])
if self.use_batchnorm:
dx, dw, db, dgamma, dbeta = affine_bn_relu_backward(dhout, ar_cache[lay])
else:
dx, dw, db = affine_relu_backward(dhout, ar_cache[lay])
grads['W%d' % (lay + 1)] = dw + self.reg * self.params['W%d' % (lay + 1)]
grads['b%d' % (lay + 1)] = db
if self.use_batchnorm:
grads['gamma%d' % (lay + 1)] = dgamma
grads['beta%d' % (lay + 1)] = dbeta
dhout = dx
# pass
############################################################################
# END OF YOUR CODE #
############################################################################
return loss, grads