import sys
sys.path.append('..')
import numpy as np
import minst as Minst
import newtowlayernet as Newtwolayernet
import preprocessing as Preprocessing
import random
import matplotlib.pylab as plt
#读入数据
train_images = Preprocessing.normalize(Minst.get_train_images())
train_lables = Preprocessing.one_hot(Minst.get_train_lables())
test_images = Preprocessing.normalize(Minst.get_test_images())
test_lables = Preprocessing.one_hot(Minst.get_test_lables())
#超参数
iters_num = 1000
train_size = train_images.shape[0]
test_size = test_images.shape[0]
batch_size = 10
learning_rate = 0.002
train_loss_list = []
train_acc_list = []
test_acc_list = []
#存放梯度 监控
w1_list = []
b1_list = []
w2_list = []
b2_list = []
'''
import sandwich as sw
import theano
theano.config.exception_verbosity = 'high'
l1 = sw.layers.InputLayer(shape=(1, 28, 28))
l2 = sw.layers.Flatten(l1)
l3 = sw.layers.FullConn(l2, n_out=500, activation=sw.activations.sigmoid)
l4 = sw.layers.FullConn(l3, n_out=10, activation=sw.activations.sigmoid)
layers = [l1, l2, l3, l4]
nnet = sw.NNet(layers)
# Load the train data
train_images = np.asarray(minst.get_train_images(), dtype='float32').reshape(60000, 1, 28, 28)
train_images[train_images<=128] = 0
train_images[train_images>128] = 1
img = train_images[0, 0]
for i in xrange(28):
print "".join(map(str, map(int, img[i])))
train_labels = minst.get_train_labels()
target = np.zeros(shape=train_labels.shape + (10, ), dtype='float32')
print target.shape
for i, label in enumerate(train_labels):
target[i, label] = 1