Re: [問題] convolution with stride VS. pooling

看板DataScience作者robert780612 (.....)時間6年前 (2018/06/26 10:58)推噓5(5推 0噓 19→)

留言24則, 8人參與討論串2/2 (看更多)

※ 引述《Haikyuu (!!)》之銘言： : convolution with stride跟pooling都是將feature map變小張的方法 : 例如convolution with stride=2的輸出大小相當於2x2的pooling的輸出大小 : 那想請問這兩者的使用上是怎麼取捨的呢? : 我認為convolution丟掉的資訊量比較少，所以DCGAN才使用convolution而非pooling : 來使Discriminator可以分辨細微差異首先，Striving for simplicity: The all convolutional net提到 'We find that max-pooling can simply be replaced by a convolutional layer with increased stride without loss in accuracy on several image recognition benchmarks' 在多個 image classification benchmark上使用convolution with stride=2替換2x2 pooling並沒有損失acc Ref: https://goo.gl/Q3c9i7 第二，用convolution with stride會比較快，以下是我使用pytorch0.4 & python3.6測試的結果使用CPU intel i5-6500 1. conv + 2*2maxpool = 0.13秒 2. conv with stride2 = 0.009秒使用GPU gtx1080 1. conv + 2*2maxpool = 0.046秒 2. conv with stride2 = 0.00069 batch size = 64 image size = 100*100 input channel = 3 output channel = 64 kernel size = 3 ### CPU import time import torch a = torch.rand(64,3,100,100) conv = torch.nn.Conv2d(3, 64, 3) pool = torch.nn.MaxPool2d(2) start = time.time() out1 = pool(conv(a)) print(time.time()-start) conv2 = torch.nn.Conv2d(3, 64, 3, stride=2) start = time.time() out2 = conv2(a) print(time.time()-start) ### GPU import time import torch import torch.nn as nn import torch.nn.functional as F class ConvPool(nn.Module): def __init__(self): super().__init__() self.conv = torch.nn.Conv2d(3, 64, 3) self.pool = torch.nn.MaxPool2d(2) def forward(self, x): out = self.pool(F.relu(self.conv(x))) return out class ConvStride(nn.Module): def __init__(self): super().__init__() self.conv = torch.nn.Conv2d(3, 64, 3, stride=2) def forward(self, x): out = F.relu(self.conv(x)) return out a = torch.rand(64,3,100,100).to('cuda') conv_pool = ConvPool().to('cuda') start = time.time() out1 = conv_pool(a) print(time.time()-start) conv_stride = ConvStride().to('cuda') start = time.time() out2 = conv_stride(a) print(time.time()-start) -- ※ 發信站: 批踢踢實業坊(ptt.cc), 來自: 211.72.69.157 ※ 文章網址: https://www.ptt.cc/bbs/DataScience/M.1529981935.A.218.html ※ 編輯: robert780612 (211.72.69.157), 06/26/2018 11:00:56 ※ 編輯: robert780612 (211.72.69.157), 06/26/2018 11:01:16

推

goldflower

06/26 14:21, 6年前 , 1^F

06/26 14:21, 1^F

※ 編輯: robert780612 (211.72.69.157), 06/26/2018 16:37:16

推

aszx4510

06/26 20:40, 6年前 , 2^F

06/26 20:40, 2^F

推

Mchord

06/26 22:41, 6年前 , 3^F

06/26 22:41, 3^F

→

Mchord

06/26 22:41, 6年前 , 4^F

06/26 22:41, 4^F

→

robert780612

06/26 23:10, 6年前 , 5^F

06/26 23:10, 5^F

推

leoloveivy

06/27 02:56, 6年前 , 6^F

06/27 02:56, 6^F

→

leoloveivy

06/27 02:56, 6年前 , 7^F

06/27 02:56, 7^F

→

leoloveivy

06/27 02:56, 6年前 , 8^F

06/27 02:56, 8^F

已補上gpu數據 ※ 編輯: robert780612 (211.72.69.157), 06/27/2018 07:53:41 ※ 編輯: robert780612 (211.72.69.157), 06/27/2018 07:54:38 ※ 編輯: robert780612 (211.72.69.157), 06/27/2018 07:55:22

→

wrt

06/28 10:31, 6年前 , 9^F

06/28 10:31, 9^F

→

wrt

06/28 10:31, 6年前 , 10^F

06/28 10:31, 10^F

→

wrt

06/28 10:31, 6年前 , 11^F

06/28 10:31, 11^F