def getResult(self, data, batch_size=100):
"""
入力データをネットワークに与え、結果を取得する。
batch_size は一度にネットワークに投げるデータ数。マシン性能により調整。
"""
self.logger.info("Get result start.")
### Model 設定
model = FunctionSet()
for num, f_layer in enumerate(self.f_layers, 1):
name = "l_f{0}".format(num)
model.__setattr__(name, f_layer)
if self.use_gpu:
model = model.to_gpu()
self.optimizer.setup(model)
### forward 処理設定
def forward(x_data):
x = Variable(x_data)
t = Variable(x_data)
h = x
for num in xrange(1, len(self.f_layers)):
h = self.activation(model.__getitem__("l_f{0}".format(num))(h))
y = model.__getitem__("l_f{0}".format(num + 1))(h)
return y.data
### 結果取得
test_data = data
test_size = len(test_data)
batch_max = int(math.ceil(test_size / float(batch_size)))
y_data = np.zeros((test_size, self.layer_sizes[len(self.layer_sizes) - 1]), dtype=test_data.dtype)
for i in xrange(batch_max):
start = i * batch_size
end = (i + 1) * batch_size
x_batch = test_data[start:end]
self.logger.debug("Index {0} => {1}, data count = {2}".format(start, end, len(x_batch)))
if self.use_gpu:
x_batch = cuda.to_gpu(x_batch)
y_batch = forward(x_batch)
if self.use_gpu:
y_batch = cuda.to_cpu(y_batch)
y_data[start:end] = y_batch
self.logger.info("Complete get result.")
return y_data
def finetuning(self, data, training_params):
"""
入力データを使用して Fine tuning を実施
"""
self.logger.info("Start finetuning...")
threshold = training_params.loss_threshold
### Model 設定
model = FunctionSet()
for num, f_layer in enumerate(self.f_layers, 1):
name = "l_f{0}".format(num)
model.__setattr__(name, f_layer)
for num, b_layer in enumerate(self.b_layers, 1):
name = "l_b{0}".format(num)
model.__setattr__(name, b_layer)
if self.use_gpu:
model = model.to_gpu()
self.optimizer.setup(model)
### forward 処理設定
def forward(x_data):
x = Variable(x_data)
t = Variable(x_data)
h = x
for num in xrange(1, len(self.f_layers) + 1):
h = self.activation(model.__getitem__("l_f{0}".format(num))(h))
for num in reversed(xrange(1, len(self.b_layers) + 1)):
h = self.activation(model.__getitem__("l_b{0}".format(num))(h))
y = h
return F.mean_squared_error(y, t)
### Training
train_data = data
train_size = len(train_data)
batch_size = training_params.batch_size
batch_max = int(math.ceil(train_size / float(batch_size)))
epoch = 0
while True:
epoch += 1
indexes = np.random.permutation(train_size)
sum_loss = 0
for i in xrange(batch_max):
start = i * batch_size
end = (i + 1) * batch_size
x_batch = train_data[indexes[start:end]]
self.logger.debug("Index {0} => {1}, data count = {2}".format(start, end, len(x_batch)))
if self.use_gpu:
x_batch = cuda.to_gpu(x_batch)
self.optimizer.zero_grads()
loss = forward(x_batch)
sum_loss += loss.data * batch_size
loss.backward()
self.optimizer.update()
epoch_loss = sum_loss / train_size
self.logger.info("Epoch {0}, Training loss = {1}".format(epoch, epoch_loss))
if threshold is not None and epoch_loss < threshold:
self.logger.info("Training loss is less then loss_threshold ({0}), stop training.".format(threshold))
break
if epoch >= training_params.epochs:
break
self.logger.info("Complete finetuning.")