#!/usr/bin/env python3 # -*- coding: UTF-8 -*- # Author: Sphantix # Mail: [email protected] # created time: 日 24 3 2019 04:57:29 下午 CST import sys, os sys.path.append(os.pardir) # 为了导入父目录而进行的设定 import numpy as np import matplotlib.pyplot as plt from mnist.mnist import load_mnist from deep_cnn_net import DeepConvNet from common.trainer import Trainer (x_train, t_train), (x_test, t_test) = load_mnist(flatten=False) network = DeepConvNet() trainer = Trainer(network, x_train, t_train, x_test, t_test, epochs=20, mini_batch_size=100, optimizer='Adam', optimizer_param={'lr': 0.001}, evaluate_sample_num_per_epoch=1000) trainer.train() # 保存参数
#!/usr/bin/env python3 # -*- coding: UTF-8 -*- # Author: Sphantix # Mail: [email protected] # created time: Wed 16 Jan 2019 12:39:51 PM CST import sys, os import numpy as np import matplotlib.pyplot as plt sys.path.append(os.pardir) from mnist.mnist import load_mnist from two_layer_net import TwoLayerNet (x_train, t_train), (x_test, t_test) = load_mnist(normalize=True, one_hot_label=True) train_loss_list = [] train_acc_list = [] test_acc_list = [] # 超参数 iters_num = 20000 train_size = x_train.shape[0] batch_size = 100 learning_rate = 0.1 # 平均每个epoch的迭代次数 iter_per_epoch = max(train_size / batch_size, 1) network = TwoLayerNet(input_size=784, hidden_size=50, output_size=10)
def get_testdata():
(x_train,t_train),(x_test,t_test) = load_mnist(flatten=True,normalize=False,one_hot_label=False)
return x_test,t_test
import numpy as np
from mnist.mnist import load_mnist
import NeuralNetwork
# データの読み込み
(x_train, t_train), (x_test, t_test) = \
load_mnist(flatten=True, normalize=True, one_hot_label=True)
# ニューラルネットワーク構築
shape = [784, 100, 100, 100, 100, 100, 10]
NN = NeuralNetwork.NeuralNetwork(shape,
batchNorm=True,
activation='tanh',
loss='CrossEntropy',
dropoutRatio=0.2)
# 学習
NN.learn(x_train,
t_train,
epoch=3000,
learningRate=0.01,
batchSize=100,
optimizer='Adam',
graph=True)
# テスト
NN.test(x_test, t_test)
# 重みファイル出力
NN.output(directory='param')
# 学習・推論処理時に利用する。 def get_normed_traindata():
