def get_act_func(config, logger):
"""This function retruns the specified activation function from the config."""
if config["activation_function"] == "ReLU":
if "ReLU" in config:
logger.debug("activation function: changed ReLu to leakyReLU with secified slope!")
return nn.LeakyReLU(negative_slope=config["ReLu"])
else:
logger.debug("activation function: ReLu")
return nn.ReLU(True)
if config["activation_function"] == "LeakyReLU":
if "LeakyReLU_negative_slope" in config:
logger.debug("activation_function: LeakyReLU")
return nn.LeakyReLU(negative_slope=config["LeakyReLU_negative_slope"])
elif "LeakyReLU" in config:
logger.debug("activation_function: LeakyReLU")
return nn.LeakyReLU(negative_slope=config["LeakyReLU"])
else:
logger.debug("activation function: LeakyReLu changed to ReLU because no slope value could be found")
return nn.LeakyReLU()
if config["activation_function"] == "Sigmoid":
logger.debug("activation_function: Sigmoid")
return nn.Sigmoid
if config["activation_function"] == "LogSigmoid":
logger.debug("activation_function: LogSigmoid")
return nn.LogSigmoid
if config["activation_function"] == "Tanh":
logger.debug("activation_function: Tanh")
return nn.Tanh
if config["activation_function"] == "SoftMax":
logger.debug("activation_function: SoftMax")
return nn.SoftMax()
def FCNN(): num_classes = 2 n_layers_enc = 32 n_layers_ctx = 128 n_input = 5 prob_drop = 0.25 layers = [] # Encoder pool = nn2.SpatialMaxPooling(2,2,2,2) layers.append(nn2.SpatialConvolution(n_input, n_layers_enc, 3, 3, 1, 1, 1, 1)) layers.append(nn.ELU()) layers.append(nn2.SpatialConvolution(n_layers_enc, n_layers_enc, 3, 3, 1, 1, 1, 1)) layers.append(nn.ELU()) layers.append(pool) # Context Module layers.append(nn2.SpatialDilatedConvolution(n_layers_enc, n_layers_ctx, 3, 3, 1, 1, 1, 1, 1, 1)) layers.append(nn.ELU()) layers.append(nn2.SpatialDropout(prob_drop)) layers.append(nn2.SpatialDilatedConvolution(n_layers_ctx, n_layers_ctx, 3, 3, 1, 1, 2, 2, 2, 2)) layers.append(nn.ELU()) layers.append(nn2.SpatialDropout(prob_drop)) layers.append(nn2.SpatialDilatedConvolution(n_layers_ctx, n_layers_ctx, 3, 3, 1, 1, 4, 4, 4, 4)) layers.append(nn.ELU()) layers.append(nn2.SpatialDropout(prob_drop)) layers.append(nn2.SpatialDilatedConvolution(n_layers_ctx, n_layers_ctx, 3, 3, 1, 1, 8, 8, 8, 8)) layers.append(nn.ELU()) layers.append(nn2.SpatialDropout(prob_drop)) layers.append(nn2.SpatialDilatedConvolution(n_layers_ctx, n_layers_ctx, 3, 3, 1, 1, 16, 16, 16, 16)) layers.append(nn.ELU()) layers.append(nn2.SpatialDropout(prob_drop)) layers.append(nn2.SpatialDilatedConvolution(n_layers_ctx, n_layers_ctx, 3, 3, 1, 1, 32, 32, 32, 32)) layers.append(nn.ELU()) layers.append(nn2.SpatialDropout(prob_drop)) layers.append(nn2.SpatialDilatedConvolution(n_layers_ctx, n_layers_ctx, 3, 3, 1, 1, 64, 64, 64, 64)) layers.append(nn.ELU()) layers.append(nn2.SpatialDropout(prob_drop)) layers.append(nn2.SpatialDilatedConvolution(n_layers_ctx, n_layers_enc, 1, 1)) layers.append(nn.ELU()) # Nao havia no paper # Decoder layers.append(nn2.SpatialMaxUnpooling(pool)) layers.append(nn2.SpatialConvolution(n_layers_enc, n_layers_enc, 3, 3, 1, 1, 1, 1)) layers.append(nn.ELU()) layers.append(nn2.SpatialConvolution(n_layers_enc, num_classes, 3, 3, 1, 1, 1, 1)) layers.append(nn.ELU()) layers.append(nn.SoftMax()) # Nao havia no paper return nn.Sequential(*layers)
def __init__(self):
super(NeuralNetwork, self).__init__()
self.features = nn.Sequential(
nn.Conv2d(1, 64, kernel_size=3, stride=2),
nn.ReLU(inplace=True),
nn.MaxPool2d(kernel_size=3, stride=2)
)
self.residual_block = nn.Sequential(
nn.Conv2d(64, 64, kernel_size=3, stride=1),
nn.ReLU(inplace=True),
nn.Conv2d(64, 64, kernel_size=3, stride=1),
nn.ReLU(inplace=True),
nn.ZeroPad2d(2)
)
self.avgpool = nn.AdaptiveAvgPool2d(output_size = (7, 7))
self.classifier = nn.Sequential(
nn.Linear(64*7*7, 2048),
nn.ReLU(inplace=True),
nn.Linear(2048, 10),
nn.SoftMax()
)
def Activation(cell_info):
if cell_info['mode'] == 'none':
return nn.Sequential()
elif cell_info['mode'] == 'tanh':
return nn.Tanh()
elif cell_info['mode'] == 'hardtanh':
return nn.Hardtanh()
elif cell_info['mode'] == 'relu':
return nn.ReLU(inplace=True)
elif cell_info['mode'] == 'prelu':
return nn.PReLU()
elif cell_info['mode'] == 'elu':
return nn.ELU(inplace=True)
elif cell_info['mode'] == 'selu':
return nn.SELU(inplace=True)
elif cell_info['mode'] == 'celu':
return nn.CELU(inplace=True)
elif cell_info['mode'] == 'sigmoid':
return nn.Sigmoid()
elif cell_info['mode'] == 'softmax':
return nn.SoftMax()
else:
raise ValueError('Not valid activation')
return
class Classification_block(self, inputChannel, outputChannel): def super(Classification_block, self).__init__(): a1 = nn.AvgPool2d(7) l1 = nn.Linear(inputChannel, outputChannel) s1 = nn.SoftMax()