文章目录
N解决了什么问题？N基本原理1）卷积层2）tf自带的2D卷积层3）动手实现卷积层Numpy版本tf2.0版本4）池化层5）动手实现池化层tf自带的MaxPool2DNumpy版本池化层tf版本池化层6）全连接层3.常用CNN架构N有哪些实际应用5.总结N解决了什么问题？
N基本原理
1）卷积层
卷积层的前向传播以及计算公式
图片:7*7*3通道:3个输入通道卷积核
输出通道高度:（5 - 3 + 2) / 2 + 1 = 3输出通道宽度:(5 - 3 + 2 ) / 2 + 1 = 3
2）tf自带的2D卷积层
3）动手实现卷积层
Numpy版本
import numpy as npdef conv_numpy(x,w,b,pad,strides):# x表示输入张量，w表示卷积核，b表示偏置，pad表示填充，strides表示步长out = None N,H.W,C = x.shapeF,HH,WW,C = w.shapeX = np.pad(x,((0,0),(pad,pad),(pad,pad),(0,0)),'constant')# 计算输出张量的高度与宽度Hn = 1 + int((H + 2 * pad - HH) / strides[0])Wn = 1 + int((W + 2 * pad - WW) / strides[1])out = np.zeros((N,Hn,Wn,F))for n in range(N):for m in range(F):for i in range(Hn):for j in range(Wn):data = X[n,i * strides[0] : i * strides[0] + HH,j * strides[1] : j * strides[1] + WW, :].reshape(1,-1) # 行filt = w[m].reshape(-1,1)# 列out[n,i,j,m] = data.dot(filt) + b(m)return out
tf2.0版本
def corr2d(x,w,b,pad,strides):# x表示输入张量，w表示卷积核，b表示偏置，pad表示填充，strides表示步长out = None N,H.W,C = tf.shape(x)F,HH,WW,C = tf.shape(w)X = tf.pad(x,((0,0),(pad,pad),(pad,pad),(0,0)),'constant')# 计算输出张量的高度与宽度Hn = 1 + int((H + 2 * pad - HH) / strides[0])Wn = 1 + int((W + 2 * pad - WW) / strides[1])Y = tf.Variable(tf.zeros((N,Hn,Wn,F),dtype = tf.float32))for m in range(F):for i in range(Hn):for j in range(Wn):data = X[:,i * strides[0] : i * 1 + HH,j * strides[1] : j * 1 + WW, :]filt = w[m,:,:,:]Y[:,i,j,m].assign(tf.reducr_sum(tf.multiply(data,filt),axis=(1,2,3)) + b[m])return Y
4）池化层
5）动手实现池化层
tf自带的MaxPool2D
Numpy版本池化层
def max_pool_forward_native(x,pool_size=(2,2),strides=(1,1)):N,H,W,C = x.shapeh_p,w_p = pool_sizeh_s,w_s = stridesHn = 1 + int((H - h_p) / h_s)Wn = 1 + ins((W - w_p) / w_s)out = np.zeros((N,Hn,Wn,C))for i in range(Hn):for j in range(Wn):out[:,i,j,:] = np.max(x[:.i*h_s : i*h_s+h_p, j*w_s： j*w_s+w_p,:],axis=(1,2))return out
tf版本池化层
def pool2d(X, pool_size=(2,2),strides=(1,1)):N,H,W,C = tf.shape(X)p_h,p_w = pool_sizes_h,s_w = stridesY = tf.zeros((N,(H - p_h + 1) // s_h, （W - p_w + 1）// s_w ,C))Y = tf.Variable(Y)for i in tf.range(tf.shape(Y)[1]):for j on tf.range(tf.shape(Y)[2]):Y[:,i,j,:].assign(tf.math.reduce_max(X[:,i*s_h : i*s_h+p_h, j*s_w : j*s_w+p_w,:],axis=(1,2),keepdims=Flase))return Y
