def __init__(self, rng, input, n_in, n_hidden, n_out):
self.hiddenLayer = HiddenLayer(
rng=rng,
input=input,
n_in=n_in,
n_out=n_hidden,
activation=T.tanh
)
self.logRegressionLayer = logisticRegression(
input=self.hiddenLayer.output,
n_in=n_hidden,
n_out=n_out
)
self.L1 = (
abs(self.hiddenLayer.W).sum()
+ abs(self.logRegressionLayer.W).sum()
)
self.L2_sqr = (
(self.hiddenLayer.W ** 2).sum()
+ (self.logRegressionLayer.W ** 2).sum()
)
self.negative_log_likelihood = (
self.logRegressionLayer.negative_log_likelihood
)
self.errors = self.logRegressionLayer.errors
self.params = self.hiddenLayer.params + self.logRegressionLayer.params
def __init__(self, rng, input, n_in, n_hidden, n_out): #多層パーセプトロンは隠れ層とロジスティック回帰で表される出力層から成る #隠れ層の出力がロジスティック回帰の入力になる点に注意 self.hiddenLayer = HiddenLayer(rng=rng, input=input, n_in=n_in, n_out=n_hidden, activation=T.tanh) self.logRegressionLayer = logisticRegression(input=self.hiddenLayer.output, n_in=n_hidden, n_out=n_out) #L1/L2正則化の正則化項を計算するシンボル self.L1 = abs(self.hiddenLayer.W).sum() + abs(self.logRegressionLayer.W).sum() self.L2_sqr = (self.hiddenLayer.W**2).sum() + (self.logRegressionLayer.W**2).sum() #MLPの誤差関数を計算するシンボル #出力層のみ依存するので ロジスティック回帰の実装と同じ self.negative_log_likelihood = self.logRegressionLayer.negative_log_likelihood #誤差を計算するシンボル self.errors = self.logRegressionLayer.errors #多層パーセプトロンのパラメータ self.params = self.hiddenLayer.params + self.logRegressionLayer.params #python的な書き方の.paramsが後に入る意味がわからない
def __init__(self, rng, input, n_in, hidden_layers_sizes, n_out):
'''
self.hiddenLayer = HiddenLayer(
rng=rng,
input=input,
n_in=n_in,
n_out=n,
activation=T.tanh
)
'''
self.params = []
self.sigmoid_layers = []
self.n_layers = len(hidden_layers_sizes)
for i in xrange(self.n_layers):
if i == 0:
input_size = n_in
else:
input_size = hidden_layers_sizes[i - 1]
if i == 0:
layer_input = input
else:
layer_input = self.sigmoid_layers[-1].output
sigmoid_layer = HiddenLayer(rng=rng,
input=layer_input,
n_in=input_size,
n_out=hidden_layers_sizes[i],
activation=T.nnet.sigmoid)
self.sigmoid_layers.append(sigmoid_layer)
self.params.extend(sigmoid_layer.params)
self.logRegressionLayer = logisticRegression(
input=self.sigmoid_layers[-1].output,
n_in=hidden_layers_sizes[-1],
n_out=n_out
)
self.L1 = abs(self.logRegressionLayer.W).sum()
for i in self.sigmoid_layers:
self.L1 += abs(i.W).sum()
'''
self.L1 = (
abs(self.hiddenLayer.W).sum()
+ abs(self.logRegressionLayer.W).sum()
)
'''
self.L2_sqr = abs(self.logRegressionLayer.W ** 2).sum()
for i in self.sigmoid_layers:
self.L2_sqr += abs(i.W ** 2).sum()
'''
self.L2_sqr = (
(self.hiddenLayer.W ** 2).sum()
+ (self.logRegressionLayer.W ** 2).sum()
)
'''
self.negative_log_likelihood = (
self.logRegressionLayer.negative_log_likelihood
)
self.errors = self.logRegressionLayer.errors
self.params.extend(self.logRegressionLayer.params)
