我正在尝试训练一个非常简单的神经网络，它有2个隐藏层（例如），每个神经网络有5个神经元。
出于某种原因，我的成本函数（选择为交叉熵，但并不重要）对于某个数据集总是在增加。
这是我的代码 -
import theano_net as N
from load import mnist
import theano
from theano import tensor as T
import numpy as np
import cv2
def floatX(X):
return np.asarray(X,dtype=theano.config.floatX)
def init_weights(shape):
return theano.shared(floatX(np.random.randn(*shape)*0.01))
def appendGD(params, grad, step):
updates = []
for p,g in zip(params,grad):
updates.append([p, p - (g * step)])
return updates
def model(X,w1,w2,wo):
h0 = X
z1 = T.dot(h0, w1.T) ## n on m1
h1 = T.nnet.sigmoid(z1)
z2 = T.dot(h1, w2.T) ## n on m2
h2 = T.nnet.sigmoid(z2)
zo = T.dot(h2, wo.T)
return T.nnet.softmax(zo)
numOfTrainPics = 4872
numOfTestPics = 382
numOfPixels = 40000
numOfLabels = 6
trX = np.zeros((numOfTrainPics,numOfPixels))
trY = np.zeros((numOfTrainPics,numOfLabels))
teX = np.zeros((numOfTestPics,numOfPixels))
teY = np.zeros((numOfTestPics,numOfLabels))
for i in range(1,4873): #generate trX and trY
img = cv2.imread('C:\\Users\\Oria\\Desktop\\Semester B\\Computer Vision Cornel 2016\\Train\\Train\\%s.jpg' %(i))
img = cv2.cvtColor(img,cv2.COLOR_RGB2GRAY)
img = np.reshape(img,(1,numOfPixels))
trX[i-1,:] = img
if(i < 1330):
trY[i-1,0] = 1
if(i > 1329)&(i < 1817):
trY[i-1,1] = 1
if(i > 1816)&(i < 2389):
trY[i-1,2] = 1
if(i > 2388)&(i < 3043):
trY[i-1,3] = 1
if(i > 3042)&(i < 4438):
trY[i-1,4] = 1
if(i > 4437)&(i < 4873):
trY[i-1,5] = 1
for i in range(1,383):
img = cv2.imread('C:\\Users\\Oria\\Desktop\\Semester B\\Computer Vision Cornel 2016\\Test\\Test\\%s.jpg' %(i))
img = cv2.cvtColor(img,cv2.COLOR_RGB2GRAY)
img = np.reshape(img,(1,numOfPixels))
teX[i-1,:] = img
if(i < 59):
teY[i-1,0] = 1
if(i > 58)&(i < 120):
teY[i-1,1] = 1
if(i > 119)&(i < 185):
teY[i-1,2] = 1
if(i > 184)&(i < 261):
teY[i-1,3] = 1
if(i > 260)&(i < 326):
teY[i-1,4] = 1
if(i > 325)&(i < 383):
teY[i-1,5] = 1
print "matrices generated"
###
x = T.fmatrix()
y = T.fmatrix()
step = 0.1
m1 = 5
m2 = 5
w1 = init_weights((m1, numOfPixels))
w2 = init_weights((m2, m1))
wo = init_weights((numOfLabels, m2))
temp = model(x, w1, w2, wo)
predictions = T.argmax(temp, axis= 1)
cost = T.mean(T.nnet.categorical_crossentropy(temp, y))
params = [w1, w2, wo]
gradient = T.grad(cost=cost, wrt = params)
update = appendGD(params, gradient, step)
train = theano.function(inputs = [x,y], outputs = cost, updates = update, allow_input_downcast=True)
predict = theano.function(inputs=[x],outputs=[predictions],allow_input_downcast=True)
for i in range(10000):
for start, end in zip(range(0,len(trX),241),range(241,len(trX),241)):
cost = train(trX[start:end], trY[start:end])
print cost
对于我正在加载此代码的trX，我的成本函数总是在增加。
然而，当我运行相同的代码但trX和trY来自MNIST数据集时，它正常工作并且成本函数正在减少。
我不明白为什么会这样，以及如何解决它。
当我看到mnist数据集的第一行trX（第一张图片）时，可能是一个线索，它是一个非常稀疏的矩阵，非零元素都在0和1之间。
[ 0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0.01176471
0.07058824 0.07058824 0.07058824 0.49411765 0.53333333 0.68627451
0.10196078 0.65098039 1. 0.96862745 0.49803922 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0.11764706 0.14117647 0.36862745
0.60392157 0.66666667 0.99215686 0.99215686 0.99215686 0.99215686
0.99215686 0.88235294 0.6745098 0.99215686 0.94901961 0.76470588
0.25098039 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0.19215686
0.93333333 0.99215686 0.99215686 0.99215686 0.99215686 0.99215686
0.99215686 0.99215686 0.99215686 0.98431373 0.36470588 0.32156863
0.32156863 0.21960784 0.15294118 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0.07058824 0.85882353 0.99215686 0.99215686 0.99215686
0.99215686 0.99215686 0.77647059 0.71372549 0.96862745 0.94509804
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0.31372549 0.61176471
0.41960784 0.99215686 0.99215686 0.80392157 0.04313725 0.
0.16862745 0.60392157 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0.05490196 0.00392157 0.60392157 0.99215686 0.35294118 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0.54509804 0.99215686 0.74509804 0.00784314
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0.04313725 0.74509804 0.99215686
0.2745098 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0.1372549
0.94509804 0.88235294 0.62745098 0.42352941 0.00392157 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0.31764706 0.94117647 0.99215686 0.99215686 0.46666667 0.09803922
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0.17647059 0.72941176 0.99215686 0.99215686
0.58823529 0.10588235 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0.0627451 0.36470588
0.98823529 0.99215686 0.73333333 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0.97647059 0.99215686 0.97647059 0.25098039 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0.18039216 0.50980392
0.71764706 0.99215686 0.99215686 0.81176471 0.00784314 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0.15294118 0.58039216 0.89803922
0.99215686 0.99215686 0.99215686 0.98039216 0.71372549 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0.09411765 0.44705882 0.86666667 0.99215686
0.99215686 0.99215686 0.99215686 0.78823529 0.30588235 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0.09019608 0.25882353 0.83529412 0.99215686 0.99215686
0.99215686 0.99215686 0.77647059 0.31764706 0.00784314 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0.07058824 0.67058824 0.85882353 0.99215686 0.99215686 0.99215686
0.99215686 0.76470588 0.31372549 0.03529412 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0.
0.21568627 0.6745098 0.88627451 0.99215686 0.99215686 0.99215686
0.99215686 0.95686275 0.52156863 0.04313725 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0.53333333 0.99215686 0.99215686 0.99215686 0.83137255
0.52941176 0.51764706 0.0627451 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. 0. 0.
0. 0. 0. 0. 0. ]
当我查看我的trX但是我的代码中加载了数据库时，trX [0]大部分都是非零，其元素在0到255之间。
我只想在我的数据库上训练神经网络。它应该不会太困难，并且代码被证明可以与MNIST一起使用。我只是不明白如何正确加载我的数据集。
答案 0 :(得分：1)
您必须将输入数据标准化为[0,1]或[-1,1]范围，因为您已经在MNIST数据集中看到了这一点。 没有规范化，训练神经网络要困难得多。
您可以通过减去数据集的平均值，除以标准偏差，或者只进行最小 - 最大标准化来实现，这将给出[0,1]范围。
对于每通道8位图像，您可以将每个像素除以255以获得[0,1]范围，这通常足以获得NN训练，但并非总是如此。