技術標籤：卷積神經網路神經網路tensorflow

簡單的卷積

from keras import layers 
from keras import models
model = models.Sequential()
model.add(layers.Conv2D(32, (3, 3), activation='relu', input_shape=(28, 28, 1))) 
model.add(layers.MaxPooling2D((2, 2)))
model.add(layers.Conv2D(64, (3, 3), activation='relu')) 
model.add(
 
layers.MaxPooling2D((2, 2)))
model.add(layers.Conv2D(64, (3, 3), activation='relu'))
model.add(layers.Flatten())
model.add(layers.Dense(64, activation='relu'))
model.add(layers.Dense(10, activation='softmax'))
model.summary()

對應的架構

Layer (type)                  Output                Shape Param #
==
 
===============================================================
conv2d_1 (Conv2D)              (None, 26, 26, 32)       320 
_________________________________________________________________
max_pooling2d_1 (MaxPooling2D) (None, 13, 13, 32)        0 
_________________________________________________________________
conv2d_2 (
 
Conv2D)              (None, 11, 11, 64)      18496 
_________________________________________________________________
max_pooling2d_2 (MaxPooling2D) (None, 5, 5, 64)          0 
_________________________________________________________________
conv2d_3 (Conv2D)              (None, 3, 3, 64)        36928
_________________________________________________________________
flatten_1 (Flatten)            (None, 576)               0 
_________________________________________________________________
dense_1 (Dense)                (None, 64)              36928 
_________________________________________________________________
dense_2 (Dense)                (None, 10)               650 
=================================================================
Total params: 93,322
Trainable params: 93,322

卷積後的輸出計算

conv2d_1 (Conv2D) 輸入為(28,28,1)，卷積核尺寸為(3,3)，過濾器個數為32，步長預設(1, 1)，所以:

Output：      (輸入尺寸-卷積核尺寸+2xpadding)/步長+1
		   	  (28-3+2x0)/1+1=26--->（26,26,32）
Shape Param： 卷積核尺寸*2 x 通道數 x 過濾器數+過濾器數
               W=3*2x1x32=288,
			   B=32,
			   Param=288+32=B+W=320

池化後的輸出計算

conv2d_1 (Conv2D)為(26,26,32)，經過池化大小為(2,2)，步長預設: If None, it will default to pool_size.

輸出尺寸： (輸入尺寸-池化尺寸)/步長+1
		   (26-2)/2+1=13---->(13,13,32)

全連線層

1.卷積後的全連線成的引數計算
全連線的前層conv2d_3 (Conv2D) (None, 3, 3, 64)
全連線層的神經元數 64

Shape Param：前層的尺寸*2x前層的核數x全連線層的神經元數+全連線層的神經元數
		     3*2 x 64 x 64 + 64= 36928

2.隱藏層的下一層全連線成的引數計算
前層神經元數：64
當前神經元數：10

Shape Param：前層神經元數 x 當前神經元數 + 當前神經元數
					64 x 10 + 10=650

Total params

為所以引數的總和：320 + 18496 +36928 + 36928 +650 = 93322