{"id":114911,"date":"2026-10-10T13:57:04","date_gmt":"2026-10-10T05:57:04","guid":{"rendered":"https:\/\/www.wsisp.com\/helps\/114911.html"},"modified":"2026-10-10T13:57:04","modified_gmt":"2026-10-10T05:57:04","slug":"cnn-%e5%8d%b7%e7%a7%af%e7%a5%9e%e7%bb%8f%e7%bd%91%e7%bb%9c%e8%af%a6%e8%a7%a3%ef%bc%9a%e4%bb%8e%e5%8e%9f%e7%90%86%e5%88%b0%e5%ae%9e%e6%88%98%ef%bc%882%ef%bc%89","status":"publish","type":"post","link":"https:\/\/www.wsisp.com\/helps\/114911.html","title":{"rendered":"CNN \u5377\u79ef\u795e\u7ecf\u7f51\u7edc\u8be6\u89e3\uff1a\u4ece\u539f\u7406\u5230\u5b9e\u6218\uff082\uff09"},"content":{"rendered":"<h3>1. \u5f15\u8a00<\/h3>\n<p>\u5377\u79ef\u795e\u7ecf\u7f51\u7edc&#xff08;Convolutional Neural Network&#xff0c;\u7b80\u79f0 CNN&#xff09;\u662f\u6df1\u5ea6\u5b66\u4e60\u9886\u57df\u6700\u5177\u5f71\u54cd\u529b\u7684\u6a21\u578b\u4e4b\u4e00&#xff0c;\u5c24\u5176\u5728\u8ba1\u7b97\u673a\u89c6\u89c9\u4efb\u52a1\u4e2d\u8868\u73b0\u5353\u8d8a\u3002\u4ece 2012 \u5e74 AlexNet \u5728 ImageNet \u7ade\u8d5b\u4e2d\u4e00\u4e3e\u593a\u51a0\u5f00\u59cb&#xff0c;CNN \u4fbf\u6210\u4e3a\u56fe\u50cf\u8bc6\u522b\u3001\u76ee\u6807\u68c0\u6d4b\u3001\u56fe\u50cf\u5206\u5272\u7b49\u4efb\u52a1\u7684\u4e3b\u6d41\u65b9\u6848\u3002\u672c\u6587\u5c06\u5e26\u4f60\u4ece\u96f6\u7406\u89e3 CNN \u7684\u6838\u5fc3\u539f\u7406&#xff0c;\u5e76\u901a\u8fc7\u4ee3\u7801\u5b9e\u6218\u638c\u63e1\u5176\u5e94\u7528\u65b9\u6cd5\u3002<\/p>\n<h3>2. \u4ec0\u4e48\u662f\u5377\u79ef\u795e\u7ecf\u7f51\u7edc<\/h3>\n<p>\u5377\u79ef\u795e\u7ecf\u7f51\u7edc\u662f\u4e00\u79cd\u4e13\u95e8\u5904\u7406\u5177\u6709\u7f51\u683c\u7ed3\u6784\u6570\u636e&#xff08;\u5982\u56fe\u50cf\u3001\u97f3\u9891&#xff09;\u7684\u6df1\u5ea6\u5b66\u4e60\u6a21\u578b\u3002\u4e0e\u4f20\u7edf\u7684\u5168\u8fde\u63a5\u795e\u7ecf\u7f51\u7edc\u4e0d\u540c&#xff0c;CNN \u901a\u8fc7\u5377\u79ef\u8fd0\u7b97\u81ea\u52a8\u63d0\u53d6\u6570\u636e\u7684\u5c40\u90e8\u7279\u5f81&#xff0c;\u5177\u6709\u5c40\u90e8\u8fde\u63a5\u548c\u6743\u503c\u5171\u4eab\u4e24\u5927\u6838\u5fc3\u7279\u6027&#xff0c;\u5927\u5e45\u51cf\u5c11\u4e86\u53c2\u6570\u91cf&#xff0c;\u63d0\u5347\u4e86\u8bad\u7ec3\u6548\u7387\u4e0e\u6cdb\u5316\u80fd\u529b\u3002<\/p>\n<h4>2.1 \u4e3a\u4ec0\u4e48\u9700\u8981 CNN<\/h4>\n<p>\u5bf9\u4e8e\u4e00\u5f20 224\u00d7224 \u7684 RGB \u56fe\u50cf&#xff0c;\u5982\u679c\u4f7f\u7528\u5168\u8fde\u63a5\u7f51\u7edc&#xff0c;\u8f93\u5165\u7ef4\u5ea6\u9ad8\u8fbe 224\u00d7224\u00d73 \u2248 15 \u4e07\u4e2a\u795e\u7ecf\u5143&#xff0c;\u7b2c\u4e00\u5c42\u5c31\u9700\u8981\u6d77\u91cf\u53c2\u6570&#xff0c;\u4e0d\u4ec5\u8ba1\u7b97\u5f00\u9500\u5de8\u5927&#xff0c;\u8fd8\u6781\u6613\u8fc7\u62df\u5408\u3002CNN \u901a\u8fc7\u5377\u79ef\u6838\u5728\u56fe\u50cf\u4e0a\u6ed1\u52a8&#xff0c;\u53ea\u5173\u6ce8\u5c40\u90e8\u533a\u57df&#xff0c;\u5e76\u7528\u540c\u4e00\u7ec4\u6743\u91cd\u626b\u63cf\u6574\u5f20\u56fe&#xff0c;\u4ece\u800c\u4ee5\u6781\u5c11\u7684\u53c2\u6570\u9ad8\u6548\u63d0\u53d6\u7279\u5f81\u3002<\/p>\n<h4>2.2 CNN \u7684\u4e09\u5927\u6838\u5fc3\u601d\u60f3<\/h4>\n<ul>\n<li>\u5c40\u90e8\u611f\u53d7\u91ce&#xff1a;\u6bcf\u4e2a\u795e\u7ecf\u5143\u53ea\u8fde\u63a5\u8f93\u5165\u7684\u4e00\u5c0f\u7247\u533a\u57df&#xff0c;\u6355\u6349\u5c40\u90e8\u7279\u5f81&#xff08;\u5982\u8fb9\u7f18\u3001\u7eb9\u7406&#xff09;\u3002<\/li>\n<li>\u6743\u503c\u5171\u4eab&#xff1a;\u540c\u4e00\u4e2a\u5377\u79ef\u6838\u5728\u6574\u5f20\u56fe\u4e0a\u6ed1\u52a8&#xff0c;\u53c2\u6570\u5171\u4eab&#xff0c;\u5927\u5e45\u51cf\u5c11\u53c2\u6570\u91cf\u3002<\/li>\n<li>\u7a7a\u95f4\u4e0b\u91c7\u6837&#xff1a;\u901a\u8fc7\u6c60\u5316\u64cd\u4f5c\u964d\u4f4e\u7279\u5f81\u56fe\u5206\u8fa8\u7387&#xff0c;\u589e\u5f3a\u5e73\u79fb\u4e0d\u53d8\u6027\u5e76\u51cf\u5c11\u8ba1\u7b97\u91cf\u3002<\/li>\n<\/ul>\n<h3>3. CNN \u7684\u6838\u5fc3\u7ed3\u6784<\/h3>\n<p>\u4e00\u4e2a\u5178\u578b\u7684 CNN \u7531\u5377\u79ef\u5c42\u3001\u6fc0\u6d3b\u51fd\u6570\u3001\u6c60\u5316\u5c42\u548c\u5168\u8fde\u63a5\u5c42\u4ea4\u66ff\u5806\u53e0\u800c\u6210\u3002<\/p>\n<h4>3.1 \u5377\u79ef\u5c42<\/h4>\n<p>\u5377\u79ef\u5c42\u662f CNN \u7684\u6838\u5fc3&#xff0c;\u901a\u8fc7\u591a\u4e2a\u5377\u79ef\u6838&#xff08;Filter&#xff09;\u5bf9\u8f93\u5165\u8fdb\u884c\u7279\u5f81\u63d0\u53d6\u3002\u6bcf\u4e2a\u5377\u79ef\u6838\u5728\u8f93\u5165\u4e0a\u6ed1\u52a8&#xff0c;\u8ba1\u7b97\u5c40\u90e8\u533a\u57df\u7684\u52a0\u6743\u548c&#xff0c;\u751f\u6210\u4e00\u5f20\u7279\u5f81\u56fe&#xff08;Feature Map&#xff09;\u3002<\/p>\n<p><span class=\"token keyword\">import<\/span> torch<br \/>\n<span class=\"token keyword\">import<\/span> torch<span class=\"token punctuation\">.<\/span>nn <span class=\"token keyword\">as<\/span> nn<\/p>\n<p><span class=\"token comment\"># \u5b9a\u4e49\u4e00\u4e2a 2D \u5377\u79ef\u5c42&#xff1a;\u8f93\u5165\u901a\u9053 3&#xff0c;\u8f93\u51fa\u901a\u9053 16&#xff0c;\u5377\u79ef\u6838 3&#215;3<\/span><br \/>\nconv_layer <span class=\"token operator\">&#061;<\/span> nn<span class=\"token punctuation\">.<\/span>Conv2d<span class=\"token punctuation\">(<\/span>in_channels<span class=\"token operator\">&#061;<\/span><span class=\"token number\">3<\/span><span class=\"token punctuation\">,<\/span> out_channels<span class=\"token operator\">&#061;<\/span><span class=\"token number\">16<\/span><span class=\"token punctuation\">,<\/span> kernel_size<span class=\"token operator\">&#061;<\/span><span class=\"token number\">3<\/span><span class=\"token punctuation\">,<\/span> stride<span class=\"token operator\">&#061;<\/span><span class=\"token number\">1<\/span><span class=\"token punctuation\">,<\/span> padding<span class=\"token operator\">&#061;<\/span><span class=\"token number\">1<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<h4>3.2 \u6fc0\u6d3b\u51fd\u6570<\/h4>\n<p>\u5377\u79ef\u662f\u7ebf\u6027\u8fd0\u7b97&#xff0c;\u9700\u8981\u5f15\u5165\u975e\u7ebf\u6027\u6fc0\u6d3b\u51fd\u6570\u6765\u589e\u5f3a\u6a21\u578b\u7684\u8868\u8fbe\u80fd\u529b\u3002\u6700\u5e38\u7528\u7684\u662f ReLU&#xff08;Rectified Linear Unit&#xff09;&#xff1a;<\/p>\n<p>relu <span class=\"token operator\">&#061;<\/span> nn<span class=\"token punctuation\">.<\/span>ReLU<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<h4>3.3 \u6c60\u5316\u5c42<\/h4>\n<p>\u6c60\u5316\u5c42\u7528\u4e8e\u4e0b\u91c7\u6837&#xff0c;\u5e38\u89c1\u7684\u6709\u6700\u5927\u6c60\u5316&#xff08;Max Pooling&#xff09;\u548c\u5e73\u5747\u6c60\u5316&#xff08;Average Pooling&#xff09;\u3002\u6700\u5927\u6c60\u5316\u53d6\u7a97\u53e3\u5185\u6700\u5927\u503c&#xff0c;\u4fdd\u7559\u6700\u663e\u8457\u7684\u7279\u5f81&#xff1a;<\/p>\n<p>pool_layer <span class=\"token operator\">&#061;<\/span> nn<span class=\"token punctuation\">.<\/span>MaxPool2d<span class=\"token punctuation\">(<\/span>kernel_size<span class=\"token operator\">&#061;<\/span><span class=\"token number\">2<\/span><span class=\"token punctuation\">,<\/span> stride<span class=\"token operator\">&#061;<\/span><span class=\"token number\">2<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<h4>3.4 \u5168\u8fde\u63a5\u5c42<\/h4>\n<p>\u7ecf\u8fc7\u591a\u5c42\u5377\u79ef\u548c\u6c60\u5316\u540e&#xff0c;\u7279\u5f81\u56fe\u88ab\u5c55\u5e73\u4e3a\u4e00\u7ef4\u5411\u91cf&#xff0c;\u9001\u5165\u5168\u8fde\u63a5\u5c42\u8fdb\u884c\u5206\u7c7b\u6216\u56de\u5f52&#xff1a;<\/p>\n<p>fc_layer <span class=\"token operator\">&#061;<\/span> nn<span class=\"token punctuation\">.<\/span>Linear<span class=\"token punctuation\">(<\/span>in_features<span class=\"token operator\">&#061;<\/span><span class=\"token number\">16<\/span> <span class=\"token operator\">*<\/span> <span class=\"token number\">56<\/span> <span class=\"token operator\">*<\/span> <span class=\"token number\">56<\/span><span class=\"token punctuation\">,<\/span> out_features<span class=\"token operator\">&#061;<\/span><span class=\"token number\">10<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<h3>4. CNN \u7684\u5b8c\u6574\u6d41\u7a0b<\/h3>\n<p>\u4e0b\u9762\u7528 Mermaid \u6d41\u7a0b\u56fe\u5c55\u793a CNN \u5904\u7406\u4e00\u5f20\u56fe\u50cf\u7684\u5b8c\u6574\u6d41\u7a0b&#xff1a;<\/p>\n<p>#mermaid-svg-nOCWXsiSIt2uADKB{font-family:\\&#8221;trebuchet ms\\&#8221;,verdana,arial,sans-serif;font-size:16px;fill:#333;}@keyframes edge-animation-frame{from{stroke-dashoffset:0;}}@keyframes dash{to{stroke-dashoffset:0;}}#mermaid-svg-nOCWXsiSIt2uADKB .edge-animation-slow{stroke-dasharray:9,5!important;stroke-dashoffset:900;animation:dash 50s linear infinite;stroke-linecap:round;}#mermaid-svg-nOCWXsiSIt2uADKB .edge-animation-fast{stroke-dasharray:9,5!important;stroke-dashoffset:900;animation:dash 20s linear infinite;stroke-linecap:round;}#mermaid-svg-nOCWXsiSIt2uADKB .error-icon{fill:#552222;}#mermaid-svg-nOCWXsiSIt2uADKB .error-text{fill:#552222;stroke:#552222;}#mermaid-svg-nOCWXsiSIt2uADKB .edge-thickness-normal{stroke-width:1px;}#mermaid-svg-nOCWXsiSIt2uADKB .edge-thickness-thick{stroke-width:3.5px;}#mermaid-svg-nOCWXsiSIt2uADKB .edge-pattern-solid{stroke-dasharray:0;}#mermaid-svg-nOCWXsiSIt2uADKB .edge-thickness-invisible{stroke-width:0;fill:none;}#mermaid-svg-nOCWXsiSIt2uADKB .edge-pattern-dashed{stroke-dasharray:3;}#mermaid-svg-nOCWXsiSIt2uADKB .edge-pattern-dotted{stroke-dasharray:2;}#mermaid-svg-nOCWXsiSIt2uADKB .marker{fill:#333333;stroke:#333333;}#mermaid-svg-nOCWXsiSIt2uADKB .marker.cross{stroke:#333333;}#mermaid-svg-nOCWXsiSIt2uADKB svg{font-family:\\&#8221;trebuchet ms\\&#8221;,verdana,arial,sans-serif;font-size:16px;}#mermaid-svg-nOCWXsiSIt2uADKB p{margin:0;}#mermaid-svg-nOCWXsiSIt2uADKB .label{font-family:\\&#8221;trebuchet ms\\&#8221;,verdana,arial,sans-serif;color:#333;}#mermaid-svg-nOCWXsiSIt2uADKB .cluster-label text{fill:#333;}#mermaid-svg-nOCWXsiSIt2uADKB .cluster-label span{color:#333;}#mermaid-svg-nOCWXsiSIt2uADKB .cluster-label span p{background-color:transparent;}#mermaid-svg-nOCWXsiSIt2uADKB .label text,#mermaid-svg-nOCWXsiSIt2uADKB span{fill:#333;color:#333;}#mermaid-svg-nOCWXsiSIt2uADKB .node rect,#mermaid-svg-nOCWXsiSIt2uADKB .node circle,#mermaid-svg-nOCWXsiSIt2uADKB .node ellipse,#mermaid-svg-nOCWXsiSIt2uADKB .node polygon,#mermaid-svg-nOCWXsiSIt2uADKB .node path{fill:#ECECFF;stroke:#9370DB;stroke-width:1px;}#mermaid-svg-nOCWXsiSIt2uADKB .rough-node .label text,#mermaid-svg-nOCWXsiSIt2uADKB .node .label text,#mermaid-svg-nOCWXsiSIt2uADKB .image-shape .label,#mermaid-svg-nOCWXsiSIt2uADKB .icon-shape .label{text-anchor:middle;}#mermaid-svg-nOCWXsiSIt2uADKB .node .katex path{fill:#000;stroke:#000;stroke-width:1px;}#mermaid-svg-nOCWXsiSIt2uADKB .rough-node .label,#mermaid-svg-nOCWXsiSIt2uADKB .node .label,#mermaid-svg-nOCWXsiSIt2uADKB .image-shape .label,#mermaid-svg-nOCWXsiSIt2uADKB .icon-shape .label{text-align:center;}#mermaid-svg-nOCWXsiSIt2uADKB .node.clickable{cursor:pointer;}#mermaid-svg-nOCWXsiSIt2uADKB .root .anchor path{fill:#333333!important;stroke-width:0;stroke:#333333;}#mermaid-svg-nOCWXsiSIt2uADKB .arrowheadPath{fill:#333333;}#mermaid-svg-nOCWXsiSIt2uADKB .edgePath .path{stroke:#333333;stroke-width:2.0px;}#mermaid-svg-nOCWXsiSIt2uADKB .flowchart-link{stroke:#333333;fill:none;}#mermaid-svg-nOCWXsiSIt2uADKB .edgeLabel{background-color:rgba(232,232,232, 0.8);text-align:center;}#mermaid-svg-nOCWXsiSIt2uADKB .edgeLabel p{background-color:rgba(232,232,232, 0.8);}#mermaid-svg-nOCWXsiSIt2uADKB .edgeLabel rect{opacity:0.5;background-color:rgba(232,232,232, 0.8);fill:rgba(232,232,232, 0.8);}#mermaid-svg-nOCWXsiSIt2uADKB .labelBkg{background-color:rgba(232, 232, 232, 0.5);}#mermaid-svg-nOCWXsiSIt2uADKB .cluster rect{fill:#ffffde;stroke:#aaaa33;stroke-width:1px;}#mermaid-svg-nOCWXsiSIt2uADKB .cluster text{fill:#333;}#mermaid-svg-nOCWXsiSIt2uADKB .cluster span{color:#333;}#mermaid-svg-nOCWXsiSIt2uADKB div.mermaidTooltip{position:absolute;text-align:center;max-width:200px;padding:2px;font-family:\\&#8221;trebuchet ms\\&#8221;,verdana,arial,sans-serif;font-size:12px;background:hsl(80, 100%, 96.2745098039%);border:1px solid #aaaa33;border-radius:2px;pointer-events:none;z-index:100;}#mermaid-svg-nOCWXsiSIt2uADKB .flowchartTitleText{text-anchor:middle;font-size:18px;fill:#333;}#mermaid-svg-nOCWXsiSIt2uADKB rect.text{fill:none;stroke-width:0;}#mermaid-svg-nOCWXsiSIt2uADKB .icon-shape,#mermaid-svg-nOCWXsiSIt2uADKB .image-shape{background-color:rgba(232,232,232, 0.8);text-align:center;}#mermaid-svg-nOCWXsiSIt2uADKB .icon-shape p,#mermaid-svg-nOCWXsiSIt2uADKB .image-shape p{background-color:rgba(232,232,232, 0.8);padding:2px;}#mermaid-svg-nOCWXsiSIt2uADKB .icon-shape .label rect,#mermaid-svg-nOCWXsiSIt2uADKB .image-shape .label rect{opacity:0.5;background-color:rgba(232,232,232, 0.8);fill:rgba(232,232,232, 0.8);}#mermaid-svg-nOCWXsiSIt2uADKB .label-icon{display:inline-block;height:1em;overflow:visible;vertical-align:-0.125em;}#mermaid-svg-nOCWXsiSIt2uADKB .node .label-icon path{fill:currentColor;stroke:revert;stroke-width:revert;}#mermaid-svg-nOCWXsiSIt2uADKB :root{&#8211;mermaid-font-family:\\&#8221;trebuchet ms\\&#8221;,verdana,arial,sans-serif;}<span class=\"edgeLabel\"><\/span><span class=\"edgeLabel\"><\/span><span class=\"edgeLabel\"><\/span><span class=\"edgeLabel\"><\/span><span class=\"edgeLabel\"><\/span><span class=\"edgeLabel\"><\/span><span class=\"edgeLabel\"><\/span><span class=\"nodeLabel\"><\/p>\n<p>\u8f93\u5165\u56fe\u50cf 224x224x3<\/p>\n<p><\/span><span class=\"nodeLabel\"><\/p>\n<p>\u5377\u79ef\u5c42 &#043; ReLU<\/p>\n<p><\/span><span class=\"nodeLabel\"><\/p>\n<p>\u6c60\u5316\u5c42<\/p>\n<p><\/span><span class=\"nodeLabel\"><\/p>\n<p>\u5377\u79ef\u5c42 &#043; ReLU<\/p>\n<p><\/span><span class=\"nodeLabel\"><\/p>\n<p>\u6c60\u5316\u5c42<\/p>\n<p><\/span><span class=\"nodeLabel\"><\/p>\n<p>\u5c55\u5e73 Flatten<\/p>\n<p><\/span><span class=\"nodeLabel\"><\/p>\n<p>\u5168\u8fde\u63a5\u5c42<\/p>\n<p><\/span><span class=\"nodeLabel\"><\/p>\n<p>Softmax \u5206\u7c7b\u8f93\u51fa<\/p>\n<p><\/span><\/p>\n<h3>5. \u5b9e\u6218&#xff1a;\u7528 PyTorch \u642d\u5efa CNN \u8bc6\u522b\u624b\u5199\u6570\u5b57<\/h3>\n<p>\u4e0b\u9762\u6211\u4eec\u4f7f\u7528 PyTorch \u5728 MNIST \u6570\u636e\u96c6\u4e0a\u642d\u5efa\u5e76\u8bad\u7ec3\u4e00\u4e2a\u7b80\u5355\u7684 CNN \u6a21\u578b\u3002<\/p>\n<h4>5.1 \u5bfc\u5165\u4f9d\u8d56\u4e0e\u52a0\u8f7d\u6570\u636e<\/h4>\n<p><span class=\"token keyword\">import<\/span> torch<br \/>\n<span class=\"token keyword\">import<\/span> torch<span class=\"token punctuation\">.<\/span>nn <span class=\"token keyword\">as<\/span> nn<br \/>\n<span class=\"token keyword\">import<\/span> torch<span class=\"token punctuation\">.<\/span>optim <span class=\"token keyword\">as<\/span> optim<br \/>\n<span class=\"token keyword\">from<\/span> torchvision <span class=\"token keyword\">import<\/span> datasets<span class=\"token punctuation\">,<\/span> transforms<br \/>\n<span class=\"token keyword\">from<\/span> torch<span class=\"token punctuation\">.<\/span>utils<span class=\"token punctuation\">.<\/span>data <span class=\"token keyword\">import<\/span> DataLoader<\/p>\n<p>transform <span class=\"token operator\">&#061;<\/span> transforms<span class=\"token punctuation\">.<\/span>Compose<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">[<\/span><br \/>\n    transforms<span class=\"token punctuation\">.<\/span>ToTensor<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">,<\/span><br \/>\n    transforms<span class=\"token punctuation\">.<\/span>Normalize<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">(<\/span><span class=\"token number\">0.1307<\/span><span class=\"token punctuation\">,<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token punctuation\">(<\/span><span class=\"token number\">0.3081<\/span><span class=\"token punctuation\">,<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token punctuation\">]<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p>train_dataset <span class=\"token operator\">&#061;<\/span> datasets<span class=\"token punctuation\">.<\/span>MNIST<span class=\"token punctuation\">(<\/span>root<span class=\"token operator\">&#061;<\/span><span class=\"token string\">&#039;.\/data&#039;<\/span><span class=\"token punctuation\">,<\/span> train<span class=\"token operator\">&#061;<\/span><span class=\"token boolean\">True<\/span><span class=\"token punctuation\">,<\/span> download<span class=\"token operator\">&#061;<\/span><span class=\"token boolean\">True<\/span><span class=\"token punctuation\">,<\/span> transform<span class=\"token operator\">&#061;<\/span>transform<span class=\"token punctuation\">)<\/span><br \/>\ntest_dataset <span class=\"token operator\">&#061;<\/span> datasets<span class=\"token punctuation\">.<\/span>MNIST<span class=\"token punctuation\">(<\/span>root<span class=\"token operator\">&#061;<\/span><span class=\"token string\">&#039;.\/data&#039;<\/span><span class=\"token punctuation\">,<\/span> train<span class=\"token operator\">&#061;<\/span><span class=\"token boolean\">False<\/span><span class=\"token punctuation\">,<\/span> download<span class=\"token operator\">&#061;<\/span><span class=\"token boolean\">True<\/span><span class=\"token punctuation\">,<\/span> transform<span class=\"token operator\">&#061;<\/span>transform<span class=\"token punctuation\">)<\/span><\/p>\n<p>train_loader <span class=\"token operator\">&#061;<\/span> DataLoader<span class=\"token punctuation\">(<\/span>train_dataset<span class=\"token punctuation\">,<\/span> batch_size<span class=\"token operator\">&#061;<\/span><span class=\"token number\">64<\/span><span class=\"token punctuation\">,<\/span> shuffle<span class=\"token operator\">&#061;<\/span><span class=\"token boolean\">True<\/span><span class=\"token punctuation\">)<\/span><br \/>\ntest_loader <span class=\"token operator\">&#061;<\/span> DataLoader<span class=\"token punctuation\">(<\/span>test_dataset<span class=\"token punctuation\">,<\/span> batch_size<span class=\"token operator\">&#061;<\/span><span class=\"token number\">64<\/span><span class=\"token punctuation\">,<\/span> shuffle<span class=\"token operator\">&#061;<\/span><span class=\"token boolean\">False<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<h4>5.2 \u5b9a\u4e49 CNN \u6a21\u578b<\/h4>\n<p><span class=\"token keyword\">class<\/span> <span class=\"token class-name\">SimpleCNN<\/span><span class=\"token punctuation\">(<\/span>nn<span class=\"token punctuation\">.<\/span>Module<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">:<\/span><br \/>\n    <span class=\"token keyword\">def<\/span> <span class=\"token function\">__init__<\/span><span class=\"token punctuation\">(<\/span>self<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">:<\/span><br \/>\n        <span class=\"token builtin\">super<\/span><span class=\"token punctuation\">(<\/span>SimpleCNN<span class=\"token punctuation\">,<\/span> self<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">.<\/span>__init__<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><br \/>\n        self<span class=\"token punctuation\">.<\/span>conv1 <span class=\"token operator\">&#061;<\/span> nn<span class=\"token punctuation\">.<\/span>Conv2d<span class=\"token punctuation\">(<\/span><span class=\"token number\">1<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token number\">32<\/span><span class=\"token punctuation\">,<\/span> kernel_size<span class=\"token operator\">&#061;<\/span><span class=\"token number\">3<\/span><span class=\"token punctuation\">,<\/span> padding<span class=\"token operator\">&#061;<\/span><span class=\"token number\">1<\/span><span class=\"token punctuation\">)<\/span><br \/>\n        self<span class=\"token punctuation\">.<\/span>conv2 <span class=\"token operator\">&#061;<\/span> nn<span class=\"token punctuation\">.<\/span>Conv2d<span class=\"token punctuation\">(<\/span><span class=\"token number\">32<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token number\">64<\/span><span class=\"token punctuation\">,<\/span> kernel_size<span class=\"token operator\">&#061;<\/span><span class=\"token number\">3<\/span><span class=\"token punctuation\">,<\/span> padding<span class=\"token operator\">&#061;<\/span><span class=\"token number\">1<\/span><span class=\"token punctuation\">)<\/span><br \/>\n        self<span class=\"token punctuation\">.<\/span>pool <span class=\"token operator\">&#061;<\/span> nn<span class=\"token punctuation\">.<\/span>MaxPool2d<span class=\"token punctuation\">(<\/span><span class=\"token number\">2<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token number\">2<\/span><span class=\"token punctuation\">)<\/span><br \/>\n        self<span class=\"token punctuation\">.<\/span>fc1 <span class=\"token operator\">&#061;<\/span> nn<span class=\"token punctuation\">.<\/span>Linear<span class=\"token punctuation\">(<\/span><span class=\"token number\">64<\/span> <span class=\"token operator\">*<\/span> <span class=\"token number\">7<\/span> <span class=\"token operator\">*<\/span> <span class=\"token number\">7<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token number\">128<\/span><span class=\"token punctuation\">)<\/span><br \/>\n        self<span class=\"token punctuation\">.<\/span>fc2 <span class=\"token operator\">&#061;<\/span> nn<span class=\"token punctuation\">.<\/span>Linear<span class=\"token punctuation\">(<\/span><span class=\"token number\">128<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token number\">10<\/span><span class=\"token punctuation\">)<\/span><br \/>\n        self<span class=\"token punctuation\">.<\/span>relu <span class=\"token operator\">&#061;<\/span> nn<span class=\"token punctuation\">.<\/span>ReLU<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p>    <span class=\"token keyword\">def<\/span> <span class=\"token function\">forward<\/span><span class=\"token punctuation\">(<\/span>self<span class=\"token punctuation\">,<\/span> x<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">:<\/span><br \/>\n        x <span class=\"token operator\">&#061;<\/span> self<span class=\"token punctuation\">.<\/span>pool<span class=\"token punctuation\">(<\/span>self<span class=\"token punctuation\">.<\/span>relu<span class=\"token punctuation\">(<\/span>self<span class=\"token punctuation\">.<\/span>conv1<span class=\"token punctuation\">(<\/span>x<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><br \/>\n        x <span class=\"token operator\">&#061;<\/span> self<span class=\"token punctuation\">.<\/span>pool<span class=\"token punctuation\">(<\/span>self<span class=\"token punctuation\">.<\/span>relu<span class=\"token punctuation\">(<\/span>self<span class=\"token punctuation\">.<\/span>conv2<span class=\"token punctuation\">(<\/span>x<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><br \/>\n        x <span class=\"token operator\">&#061;<\/span> x<span class=\"token punctuation\">.<\/span>view<span class=\"token punctuation\">(<\/span><span class=\"token operator\">&#8211;<\/span><span class=\"token number\">1<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token number\">64<\/span> <span class=\"token operator\">*<\/span> <span class=\"token number\">7<\/span> <span class=\"token operator\">*<\/span> <span class=\"token number\">7<\/span><span class=\"token punctuation\">)<\/span><br \/>\n        x <span class=\"token operator\">&#061;<\/span> self<span class=\"token punctuation\">.<\/span>relu<span class=\"token punctuation\">(<\/span>self<span class=\"token punctuation\">.<\/span>fc1<span class=\"token punctuation\">(<\/span>x<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><br \/>\n        x <span class=\"token operator\">&#061;<\/span> self<span class=\"token punctuation\">.<\/span>fc2<span class=\"token punctuation\">(<\/span>x<span class=\"token punctuation\">)<\/span><br \/>\n        <span class=\"token keyword\">return<\/span> x<\/p>\n<p>model <span class=\"token operator\">&#061;<\/span> SimpleCNN<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<h4>5.3 \u8bad\u7ec3\u6a21\u578b<\/h4>\n<h4>5.3.1 \u7ed8\u5236\u8bad\u7ec3\u635f\u5931\u66f2\u7ebf<\/h4>\n<p>\u4e3a\u4e86\u76f4\u89c2\u89c2\u5bdf\u6a21\u578b\u7684\u6536\u655b\u60c5\u51b5&#xff0c;\u6211\u4eec\u53ef\u4ee5\u5728\u8bad\u7ec3\u8fc7\u7a0b\u4e2d\u8bb0\u5f55\u6bcf\u4e2a epoch \u7684\u5e73\u5747\u635f\u5931&#xff0c;\u5e76\u4f7f\u7528 matplotlib \u7ed8\u5236\u635f\u5931\u968f epoch \u53d8\u5316\u7684\u66f2\u7ebf\u56fe\u3002\u9996\u5148\u9700\u8981\u5b89\u88c5\u5e76\u5bfc\u5165 matplotlib&#xff1a;<\/p>\n<p><span class=\"token keyword\">import<\/span> matplotlib<span class=\"token punctuation\">.<\/span>pyplot <span class=\"token keyword\">as<\/span> plt<\/p>\n<p>\u4fee\u6539\u8bad\u7ec3\u51fd\u6570&#xff0c;\u5728\u8bad\u7ec3\u8fc7\u7a0b\u4e2d\u8bb0\u5f55\u6bcf\u4e2a epoch \u7684\u5e73\u5747\u635f\u5931&#xff1a;<\/p>\n<p><span class=\"token keyword\">def<\/span> <span class=\"token function\">train<\/span><span class=\"token punctuation\">(<\/span>epochs<span class=\"token operator\">&#061;<\/span><span class=\"token number\">3<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">:<\/span><br \/>\n    model<span class=\"token punctuation\">.<\/span>train<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><br \/>\n    train_losses <span class=\"token operator\">&#061;<\/span> <span class=\"token punctuation\">[<\/span><span class=\"token punctuation\">]<\/span>  <span class=\"token comment\"># \u8bb0\u5f55\u6bcf\u4e2a epoch \u7684\u5e73\u5747\u635f\u5931<\/span><br \/>\n    <span class=\"token keyword\">for<\/span> epoch <span class=\"token keyword\">in<\/span> <span class=\"token builtin\">range<\/span><span class=\"token punctuation\">(<\/span>epochs<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">:<\/span><br \/>\n        total_loss <span class=\"token operator\">&#061;<\/span> <span class=\"token number\">0<\/span><br \/>\n        <span class=\"token keyword\">for<\/span> images<span class=\"token punctuation\">,<\/span> labels <span class=\"token keyword\">in<\/span> train_loader<span class=\"token punctuation\">:<\/span><br \/>\n            images<span class=\"token punctuation\">,<\/span> labels <span class=\"token operator\">&#061;<\/span> images<span class=\"token punctuation\">.<\/span>to<span class=\"token punctuation\">(<\/span>device<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">,<\/span> labels<span class=\"token punctuation\">.<\/span>to<span class=\"token punctuation\">(<\/span>device<span class=\"token punctuation\">)<\/span><br \/>\n            optimizer<span class=\"token punctuation\">.<\/span>zero_grad<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><br \/>\n            outputs <span class=\"token operator\">&#061;<\/span> model<span class=\"token punctuation\">(<\/span>images<span class=\"token punctuation\">)<\/span><br \/>\n            loss <span class=\"token operator\">&#061;<\/span> criterion<span class=\"token punctuation\">(<\/span>outputs<span class=\"token punctuation\">,<\/span> labels<span class=\"token punctuation\">)<\/span><br \/>\n            loss<span class=\"token punctuation\">.<\/span>backward<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><br \/>\n            optimizer<span class=\"token punctuation\">.<\/span>step<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><br \/>\n            total_loss <span class=\"token operator\">&#043;&#061;<\/span> loss<span class=\"token punctuation\">.<\/span>item<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><br \/>\n        avg_loss <span class=\"token operator\">&#061;<\/span> total_loss <span class=\"token operator\">\/<\/span> <span class=\"token builtin\">len<\/span><span class=\"token punctuation\">(<\/span>train_loader<span class=\"token punctuation\">)<\/span><br \/>\n        train_losses<span class=\"token punctuation\">.<\/span>append<span class=\"token punctuation\">(<\/span>avg_loss<span class=\"token punctuation\">)<\/span><br \/>\n        <span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string-interpolation\"><span class=\"token string\">f&#039;Epoch <\/span><span class=\"token interpolation\"><span class=\"token punctuation\">{<\/span>epoch<span class=\"token operator\">&#043;<\/span><span class=\"token number\">1<\/span><span class=\"token punctuation\">}<\/span><\/span><span class=\"token string\">, Loss: <\/span><span class=\"token interpolation\"><span class=\"token punctuation\">{<\/span>avg_loss<span class=\"token punctuation\">:<\/span><span class=\"token format-spec\">.4f<\/span><span class=\"token punctuation\">}<\/span><\/span><span class=\"token string\">&#039;<\/span><\/span><span class=\"token punctuation\">)<\/span><br \/>\n    <span class=\"token keyword\">return<\/span> train_losses<\/p>\n<p>train_losses <span class=\"token operator\">&#061;<\/span> train<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p>\u8bad\u7ec3\u5b8c\u6210\u540e&#xff0c;\u4f7f\u7528 matplotlib \u7ed8\u5236\u635f\u5931\u66f2\u7ebf&#xff1a;<\/p>\n<p>plt<span class=\"token punctuation\">.<\/span>figure<span class=\"token punctuation\">(<\/span>figsize<span class=\"token operator\">&#061;<\/span><span class=\"token punctuation\">(<\/span><span class=\"token number\">8<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token number\">5<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><br \/>\nplt<span class=\"token punctuation\">.<\/span>plot<span class=\"token punctuation\">(<\/span><span class=\"token builtin\">range<\/span><span class=\"token punctuation\">(<\/span><span class=\"token number\">1<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token builtin\">len<\/span><span class=\"token punctuation\">(<\/span>train_losses<span class=\"token punctuation\">)<\/span> <span class=\"token operator\">&#043;<\/span> <span class=\"token number\">1<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">,<\/span> train_losses<span class=\"token punctuation\">,<\/span> marker<span class=\"token operator\">&#061;<\/span><span class=\"token string\">&#039;o&#039;<\/span><span class=\"token punctuation\">,<\/span> linestyle<span class=\"token operator\">&#061;<\/span><span class=\"token string\">&#039;-&#039;<\/span><span class=\"token punctuation\">,<\/span> color<span class=\"token operator\">&#061;<\/span><span class=\"token string\">&#039;b&#039;<\/span><span class=\"token punctuation\">)<\/span><br \/>\nplt<span class=\"token punctuation\">.<\/span>title<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#039;Training Loss Curve&#039;<\/span><span class=\"token punctuation\">)<\/span><br \/>\nplt<span class=\"token punctuation\">.<\/span>xlabel<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#039;Epoch&#039;<\/span><span class=\"token punctuation\">)<\/span><br \/>\nplt<span class=\"token punctuation\">.<\/span>ylabel<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#039;Loss&#039;<\/span><span class=\"token punctuation\">)<\/span><br \/>\nplt<span class=\"token punctuation\">.<\/span>xticks<span class=\"token punctuation\">(<\/span><span class=\"token builtin\">range<\/span><span class=\"token punctuation\">(<\/span><span class=\"token number\">1<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token builtin\">len<\/span><span class=\"token punctuation\">(<\/span>train_losses<span class=\"token punctuation\">)<\/span> <span class=\"token operator\">&#043;<\/span> <span class=\"token number\">1<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><br \/>\nplt<span class=\"token punctuation\">.<\/span>grid<span class=\"token punctuation\">(<\/span><span class=\"token boolean\">True<\/span><span class=\"token punctuation\">)<\/span><br \/>\nplt<span class=\"token punctuation\">.<\/span>show<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p>\u793a\u4f8b\u8f93\u51fa&#xff1a;<\/p>\n<p>Epoch 1, Loss: 0.2345<br \/>\nEpoch 2, Loss: 0.0876<br \/>\nEpoch 3, Loss: 0.0521<\/p>\n<p>\u8fd0\u884c\u4e0a\u8ff0\u4ee3\u7801\u540e&#xff0c;\u4f1a\u5f39\u51fa\u4e00\u4e2a\u7a97\u53e3&#xff0c;\u663e\u793a\u4e00\u6761\u968f epoch \u589e\u52a0\u800c\u9010\u6e10\u4e0b\u964d\u7684\u635f\u5931\u66f2\u7ebf&#xff0c;\u8bf4\u660e\u6a21\u578b\u5728\u8bad\u7ec3\u8fc7\u7a0b\u4e2d\u4e0d\u65ad\u6536\u655b\u3002\u968f\u7740\u8bad\u7ec3\u8f6e\u6570\u589e\u52a0&#xff0c;\u635f\u5931\u503c\u4f1a\u8d8a\u6765\u8d8a\u5c0f\u5e76\u8d8b\u4e8e\u5e73\u7a33&#xff0c;\u8fd9\u6b63\u662f\u6211\u4eec\u671f\u671b\u770b\u5230\u7684\u8bad\u7ec3\u6548\u679c\u3002<\/p>\n<p>device <span class=\"token operator\">&#061;<\/span> torch<span class=\"token punctuation\">.<\/span>device<span class=\"token punctuation\">(<\/span><span class=\"token string\">&#039;cuda&#039;<\/span> <span class=\"token keyword\">if<\/span> torch<span class=\"token punctuation\">.<\/span>cuda<span class=\"token punctuation\">.<\/span>is_available<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span> <span class=\"token keyword\">else<\/span> <span class=\"token string\">&#039;cpu&#039;<\/span><span class=\"token punctuation\">)<\/span><br \/>\nmodel <span class=\"token operator\">&#061;<\/span> model<span class=\"token punctuation\">.<\/span>to<span class=\"token punctuation\">(<\/span>device<span class=\"token punctuation\">)<\/span><br \/>\ncriterion <span class=\"token operator\">&#061;<\/span> nn<span class=\"token punctuation\">.<\/span>CrossEntropyLoss<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><br \/>\noptimizer <span class=\"token operator\">&#061;<\/span> optim<span class=\"token punctuation\">.<\/span>Adam<span class=\"token punctuation\">(<\/span>model<span class=\"token punctuation\">.<\/span>parameters<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">,<\/span> lr<span class=\"token operator\">&#061;<\/span><span class=\"token number\">0.001<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p><span class=\"token keyword\">def<\/span> <span class=\"token function\">train<\/span><span class=\"token punctuation\">(<\/span>epochs<span class=\"token operator\">&#061;<\/span><span class=\"token number\">3<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">:<\/span><br \/>\n    model<span class=\"token punctuation\">.<\/span>train<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><br \/>\n    <span class=\"token keyword\">for<\/span> epoch <span class=\"token keyword\">in<\/span> <span class=\"token builtin\">range<\/span><span class=\"token punctuation\">(<\/span>epochs<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">:<\/span><br \/>\n        total_loss <span class=\"token operator\">&#061;<\/span> <span class=\"token number\">0<\/span><br \/>\n        <span class=\"token keyword\">for<\/span> images<span class=\"token punctuation\">,<\/span> labels <span class=\"token keyword\">in<\/span> train_loader<span class=\"token punctuation\">:<\/span><br \/>\n            images<span class=\"token punctuation\">,<\/span> labels <span class=\"token operator\">&#061;<\/span> images<span class=\"token punctuation\">.<\/span>to<span class=\"token punctuation\">(<\/span>device<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">,<\/span> labels<span class=\"token punctuation\">.<\/span>to<span class=\"token punctuation\">(<\/span>device<span class=\"token punctuation\">)<\/span><br \/>\n            optimizer<span class=\"token punctuation\">.<\/span>zero_grad<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><br \/>\n            outputs <span class=\"token operator\">&#061;<\/span> model<span class=\"token punctuation\">(<\/span>images<span class=\"token punctuation\">)<\/span><br \/>\n            loss <span class=\"token operator\">&#061;<\/span> criterion<span class=\"token punctuation\">(<\/span>outputs<span class=\"token punctuation\">,<\/span> labels<span class=\"token punctuation\">)<\/span><br \/>\n            loss<span class=\"token punctuation\">.<\/span>backward<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><br \/>\n            optimizer<span class=\"token punctuation\">.<\/span>step<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><br \/>\n            total_loss <span class=\"token operator\">&#043;&#061;<\/span> loss<span class=\"token punctuation\">.<\/span>item<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><br \/>\n        <span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string-interpolation\"><span class=\"token string\">f&#039;Epoch <\/span><span class=\"token interpolation\"><span class=\"token punctuation\">{<\/span>epoch<span class=\"token operator\">&#043;<\/span><span class=\"token number\">1<\/span><span class=\"token punctuation\">}<\/span><\/span><span class=\"token string\">, Loss: <\/span><span class=\"token interpolation\"><span class=\"token punctuation\">{<\/span>total_loss <span class=\"token operator\">\/<\/span> <span class=\"token builtin\">len<\/span><span class=\"token punctuation\">(<\/span>train_loader<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">:<\/span><span class=\"token format-spec\">.4f<\/span><span class=\"token punctuation\">}<\/span><\/span><span class=\"token string\">&#039;<\/span><\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p>train<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<h4>5.4 \u8bc4\u4f30\u6a21\u578b<\/h4>\n<p><span class=\"token keyword\">def<\/span> <span class=\"token function\">evaluate<\/span><span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">:<\/span><br \/>\n    model<span class=\"token punctuation\">.<\/span><span class=\"token builtin\">eval<\/span><span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><br \/>\n    correct <span class=\"token operator\">&#061;<\/span> <span class=\"token number\">0<\/span><br \/>\n    total <span class=\"token operator\">&#061;<\/span> <span class=\"token number\">0<\/span><br \/>\n    <span class=\"token keyword\">with<\/span> torch<span class=\"token punctuation\">.<\/span>no_grad<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">:<\/span><br \/>\n        <span class=\"token keyword\">for<\/span> images<span class=\"token punctuation\">,<\/span> labels <span class=\"token keyword\">in<\/span> test_loader<span class=\"token punctuation\">:<\/span><br \/>\n            images<span class=\"token punctuation\">,<\/span> labels <span class=\"token operator\">&#061;<\/span> images<span class=\"token punctuation\">.<\/span>to<span class=\"token punctuation\">(<\/span>device<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">,<\/span> labels<span class=\"token punctuation\">.<\/span>to<span class=\"token punctuation\">(<\/span>device<span class=\"token punctuation\">)<\/span><br \/>\n            outputs <span class=\"token operator\">&#061;<\/span> model<span class=\"token punctuation\">(<\/span>images<span class=\"token punctuation\">)<\/span><br \/>\n            _<span class=\"token punctuation\">,<\/span> predicted <span class=\"token operator\">&#061;<\/span> torch<span class=\"token punctuation\">.<\/span><span class=\"token builtin\">max<\/span><span class=\"token punctuation\">(<\/span>outputs<span class=\"token punctuation\">,<\/span> <span class=\"token number\">1<\/span><span class=\"token punctuation\">)<\/span><br \/>\n            total <span class=\"token operator\">&#043;&#061;<\/span> labels<span class=\"token punctuation\">.<\/span>size<span class=\"token punctuation\">(<\/span><span class=\"token number\">0<\/span><span class=\"token punctuation\">)<\/span><br \/>\n            correct <span class=\"token operator\">&#043;&#061;<\/span> <span class=\"token punctuation\">(<\/span>predicted <span class=\"token operator\">&#061;&#061;<\/span> labels<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">.<\/span><span class=\"token builtin\">sum<\/span><span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">.<\/span>item<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><br \/>\n    <span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string-interpolation\"><span class=\"token string\">f&#039;Accuracy: <\/span><span class=\"token interpolation\"><span class=\"token punctuation\">{<\/span><span class=\"token number\">100<\/span> <span class=\"token operator\">*<\/span> correct <span class=\"token operator\">\/<\/span> total<span class=\"token punctuation\">:<\/span><span class=\"token format-spec\">.2f<\/span><span class=\"token punctuation\">}<\/span><\/span><span class=\"token string\">%&#039;<\/span><\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p>evaluate<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<h3>6. \u7ecf\u5178 CNN \u67b6\u6784\u6f14\u8fdb<\/h3>\n<p>\u4e86\u89e3\u7ecf\u5178\u7f51\u7edc\u7ed3\u6784\u6709\u52a9\u4e8e\u7406\u89e3 CNN \u7684\u53d1\u5c55\u8109\u7edc&#xff1a;<\/p>\n<ul>\n<li>LeNet-5&#xff08;1998&#xff09;&#xff1a;CNN \u7684\u5f00\u5c71\u4e4b\u4f5c&#xff0c;\u7528\u4e8e\u624b\u5199\u6570\u5b57\u8bc6\u522b\u3002<\/li>\n<li>AlexNet&#xff08;2012&#xff09;&#xff1a;\u5f15\u5165 ReLU\u3001Dropout \u548c\u6570\u636e\u589e\u5f3a&#xff0c;\u5728 ImageNet \u4e0a\u5927\u5e45\u9886\u5148\u3002<\/li>\n<li>VGGNet&#xff08;2014&#xff09;&#xff1a;\u4f7f\u7528\u5c0f\u5377\u79ef\u6838\u5806\u53e0\u66f4\u6df1\u7f51\u7edc&#xff0c;\u7ed3\u6784\u89c4\u6574\u3002<\/li>\n<li>GoogLeNet&#xff08;2014&#xff09;&#xff1a;\u5f15\u5165 Inception \u6a21\u5757&#xff0c;\u63d0\u5347\u5bbd\u5ea6\u3002<\/li>\n<li>ResNet&#xff08;2015&#xff09;&#xff1a;\u5f15\u5165\u6b8b\u5dee\u8fde\u63a5&#xff0c;\u89e3\u51b3\u6df1\u5c42\u7f51\u7edc\u9000\u5316\u95ee\u9898\u3002<\/li>\n<\/ul>\n<p>\u4e0b\u8868\u4ece\u63d0\u51fa\u5e74\u4efd\u3001\u6838\u5fc3\u521b\u65b0\u70b9\u3001\u53c2\u6570\u91cf\u7ea7\u548c\u4e3b\u8981\u5e94\u7528\u573a\u666f\u56db\u4e2a\u7ef4\u5ea6&#xff0c;\u5bf9\u8fd9\u4e94\u4e2a\u7ecf\u5178\u67b6\u6784\u8fdb\u884c\u6a2a\u5411\u5bf9\u6bd4&#xff1a;<\/p>\n<table>\n<tr>\u6a21\u578b\u63d0\u51fa\u5e74\u4efd\u6838\u5fc3\u521b\u65b0\u70b9\u53c2\u6570\u91cf\u7ea7\u4e3b\u8981\u5e94\u7528\u573a\u666f<\/tr>\n<tbody>\n<tr>\n<td>LeNet-5<\/td>\n<td>1998<\/td>\n<td>\u9996\u4e2a\u6210\u529f\u5e94\u7528\u7684 CNN&#xff0c;\u63d0\u51fa\u5377\u79ef &#043; \u6c60\u5316 &#043; \u5168\u8fde\u63a5\u7684\u6807\u51c6\u7ed3\u6784<\/td>\n<td>\u7ea6 6 \u4e07<\/td>\n<td>\u624b\u5199\u6570\u5b57\u8bc6\u522b\u3001\u652f\u7968\/\u90ae\u7f16\u8bfb\u53d6<\/td>\n<\/tr>\n<tr>\n<td>AlexNet<\/td>\n<td>2012<\/td>\n<td>\u5f15\u5165 ReLU\u3001Dropout\u3001\u6570\u636e\u589e\u5f3a\u4e0e GPU \u5e76\u884c\u8bad\u7ec3&#xff0c;\u5927\u5e45\u52a0\u6df1\u7f51\u7edc<\/td>\n<td>\u7ea6 6000 \u4e07<\/td>\n<td>ImageNet \u5927\u89c4\u6a21\u56fe\u50cf\u5206\u7c7b<\/td>\n<\/tr>\n<tr>\n<td>VGGNet<\/td>\n<td>2014<\/td>\n<td>\u4f7f\u7528\u591a\u4e2a 3\u00d73 \u5c0f\u5377\u79ef\u6838\u5806\u53e0&#xff0c;\u7ed3\u6784\u89c4\u6574\u3001\u6613\u4e8e\u6269\u5c55<\/td>\n<td>\u7ea6 1.38 \u4ebf<\/td>\n<td>\u56fe\u50cf\u5206\u7c7b\u3001\u7279\u5f81\u63d0\u53d6\u9aa8\u5e72\u7f51\u7edc<\/td>\n<\/tr>\n<tr>\n<td>GoogLeNet<\/td>\n<td>2014<\/td>\n<td>\u63d0\u51fa Inception \u6a21\u5757&#xff0c;\u5728\u589e\u52a0\u5bbd\u5ea6\u7684\u540c\u65f6\u63a7\u5236\u8ba1\u7b97\u91cf<\/td>\n<td>\u7ea6 500 \u4e07<\/td>\n<td>\u56fe\u50cf\u5206\u7c7b\u3001\u79fb\u52a8\u7aef\/\u4f4e\u7b97\u529b\u573a\u666f<\/td>\n<\/tr>\n<tr>\n<td>ResNet<\/td>\n<td>2015<\/td>\n<td>\u5f15\u5165\u6b8b\u5dee\u8fde\u63a5&#xff08;Skip Connection&#xff09;&#xff0c;\u89e3\u51b3\u6df1\u5c42\u7f51\u7edc\u9000\u5316\u95ee\u9898<\/td>\n<td>\u7ea6 2500 \u4e07&#xff08;ResNet-50&#xff09;<\/td>\n<td>\u56fe\u50cf\u5206\u7c7b\u3001\u68c0\u6d4b\u3001\u5206\u5272\u7b49\u901a\u7528\u9aa8\u5e72\u7f51\u7edc<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<h3>7. \u603b\u7ed3<\/h3>\n<p>CNN \u901a\u8fc7\u5c40\u90e8\u8fde\u63a5\u4e0e\u6743\u503c\u5171\u4eab\u5927\u5e45\u51cf\u5c11\u53c2\u6570\u91cf&#xff0c;\u4ee5\u5377\u79ef\u5c42\u63d0\u53d6\u5c40\u90e8\u7279\u5f81\u3001\u6fc0\u6d3b\u51fd\u6570\u5f15\u5165\u975e\u7ebf\u6027\u3001\u6c60\u5316\u5c42\u4e0b\u91c7\u6837\u964d\u7ef4\u3001\u5168\u8fde\u63a5\u5c42\u5b8c\u6210\u5206\u7c7b&#xff0c;\u6784\u6210\u4e86\u8ba1\u7b97\u673a\u89c6\u89c9\u7684\u57fa\u77f3\u3002\u5b9e\u6218\u4e2d&#xff0c;\u6211\u4eec\u4f7f\u7528 PyTorch \u5728 MNIST \u4e0a\u5b8c\u6210\u4e86\u6570\u636e\u52a0\u8f7d\u3001\u6a21\u578b\u5b9a\u4e49\u3001\u8bad\u7ec3\u4e0e\u8bc4\u4f30\u7684\u5b8c\u6574\u6d41\u7a0b&#xff0c;\u5e76\u501f\u52a9\u635f\u5931\u66f2\u7ebf\u76f4\u89c2\u9a8c\u8bc1\u4e86\u6a21\u578b\u7684\u6536\u655b\u8fc7\u7a0b\u3002\u638c\u63e1\u8fd9\u4e9b\u57fa\u7840\u540e&#xff0c;\u53ef\u8fdb\u4e00\u6b65\u5b66\u4e60\u76ee\u6807\u68c0\u6d4b&#xff08;YOLO\u3001Faster R-CNN&#xff09;\u3001\u56fe\u50cf\u5206\u5272&#xff08;U-Net\u3001Mask R-CNN&#xff09;\u4ee5\u53ca Transformer \u7b49\u66f4\u524d\u6cbf\u7684\u89c6\u89c9\u6a21\u578b&#xff0c;\u6301\u7eed\u62d3\u5c55\u5e94\u7528\u8fb9\u754c\u3002<\/p>\n<h3>8. CNN \u7684\u4f18\u5316\u6280\u5de7<\/h3>\n<p>\u8bad\u7ec3 CNN \u65f6&#xff0c;\u9664\u4e86\u6a21\u578b\u7ed3\u6784\u672c\u8eab&#xff0c;\u4e00\u4e9b\u8bad\u7ec3\u6280\u5de7\u5f80\u5f80\u80fd\u663e\u8457\u63d0\u5347\u6700\u7ec8\u7cbe\u5ea6\u3002\u4e0b\u9762\u603b\u7ed3\u51e0\u4e2a\u5b9e\u7528\u4e14\u5bb9\u6613\u4e0a\u624b\u7684\u4f18\u5316\u65b9\u5411\u3002<\/p>\n<h4>8.1 \u6570\u636e\u589e\u5f3a<\/h4>\n<p>\u6570\u636e\u589e\u5f3a\u901a\u8fc7\u5bf9\u8bad\u7ec3\u6837\u672c\u505a\u968f\u673a\u53d8\u6362&#xff0c;\u6269\u5145\u6570\u636e\u591a\u6837\u6027&#xff0c;\u80fd\u6709\u6548\u7f13\u89e3\u8fc7\u62df\u5408\u3002\u5e38\u7528\u7684\u589e\u5f3a\u624b\u6bb5\u5305\u62ec\u968f\u673a\u88c1\u526a\u3001\u6c34\u5e73\u7ffb\u8f6c\u3001\u65cb\u8f6c\u3001\u8272\u5f69\u6296\u52a8\u7b49&#xff1a;<\/p>\n<p><span class=\"token keyword\">from<\/span> torchvision <span class=\"token keyword\">import<\/span> transforms<\/p>\n<p>train_transform <span class=\"token operator\">&#061;<\/span> transforms<span class=\"token punctuation\">.<\/span>Compose<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">[<\/span><br \/>\n    transforms<span class=\"token punctuation\">.<\/span>RandomResizedCrop<span class=\"token punctuation\">(<\/span><span class=\"token number\">224<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">,<\/span><br \/>\n    transforms<span class=\"token punctuation\">.<\/span>RandomHorizontalFlip<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">,<\/span><br \/>\n    transforms<span class=\"token punctuation\">.<\/span>ColorJitter<span class=\"token punctuation\">(<\/span>brightness<span class=\"token operator\">&#061;<\/span><span class=\"token number\">0.2<\/span><span class=\"token punctuation\">,<\/span> contrast<span class=\"token operator\">&#061;<\/span><span class=\"token number\">0.2<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">,<\/span><br \/>\n    transforms<span class=\"token punctuation\">.<\/span>ToTensor<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">,<\/span><br \/>\n    transforms<span class=\"token punctuation\">.<\/span>Normalize<span class=\"token punctuation\">(<\/span>mean<span class=\"token operator\">&#061;<\/span><span class=\"token punctuation\">[<\/span><span class=\"token number\">0.485<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token number\">0.456<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token number\">0.406<\/span><span class=\"token punctuation\">]<\/span><span class=\"token punctuation\">,<\/span> std<span class=\"token operator\">&#061;<\/span><span class=\"token punctuation\">[<\/span><span class=\"token number\">0.229<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token number\">0.224<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token number\">0.225<\/span><span class=\"token punctuation\">]<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token punctuation\">]<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<h4>8.2 \u5b66\u4e60\u7387\u8c03\u5ea6<\/h4>\n<p>\u8bad\u7ec3\u521d\u671f\u4f7f\u7528\u8f83\u5927\u5b66\u4e60\u7387\u5feb\u901f\u6536\u655b&#xff0c;\u540e\u671f\u9010\u6b65\u964d\u4f4e\u5b66\u4e60\u7387\u4ee5\u7cbe\u7ec6\u903c\u8fd1\u6700\u4f18\u89e3\u3002PyTorch \u63d0\u4f9b\u4e86\u591a\u79cd\u8c03\u5ea6\u5668&#xff1a;<\/p>\n<p><span class=\"token keyword\">import<\/span> torch<span class=\"token punctuation\">.<\/span>optim <span class=\"token keyword\">as<\/span> optim<br \/>\n<span class=\"token keyword\">from<\/span> torch<span class=\"token punctuation\">.<\/span>optim<span class=\"token punctuation\">.<\/span>lr_scheduler <span class=\"token keyword\">import<\/span> StepLR<\/p>\n<p>optimizer <span class=\"token operator\">&#061;<\/span> optim<span class=\"token punctuation\">.<\/span>Adam<span class=\"token punctuation\">(<\/span>model<span class=\"token punctuation\">.<\/span>parameters<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">,<\/span> lr<span class=\"token operator\">&#061;<\/span><span class=\"token number\">0.001<\/span><span class=\"token punctuation\">)<\/span><br \/>\nscheduler <span class=\"token operator\">&#061;<\/span> StepLR<span class=\"token punctuation\">(<\/span>optimizer<span class=\"token punctuation\">,<\/span> step_size<span class=\"token operator\">&#061;<\/span><span class=\"token number\">5<\/span><span class=\"token punctuation\">,<\/span> gamma<span class=\"token operator\">&#061;<\/span><span class=\"token number\">0.1<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p><span class=\"token comment\"># \u6bcf\u4e2a epoch \u7ed3\u675f\u540e\u8c03\u7528<\/span><br \/>\n<span class=\"token comment\"># scheduler.step()<\/span><\/p>\n<h4>8.3 \u6b63\u5219\u5316\u4e0e Dropout<\/h4>\n<p>Dropout \u5728\u8bad\u7ec3\u65f6\u968f\u673a\u4e22\u5f03\u4e00\u90e8\u5206\u795e\u7ecf\u5143&#xff0c;\u8feb\u4f7f\u7f51\u7edc\u5b66\u4e60\u66f4\u9c81\u68d2\u7684\u7279\u5f81&#xff0c;\u662f\u9632\u6b62\u8fc7\u62df\u5408\u7684\u5e38\u7528\u624b\u6bb5&#xff1a;<\/p>\n<p>self<span class=\"token punctuation\">.<\/span>dropout <span class=\"token operator\">&#061;<\/span> nn<span class=\"token punctuation\">.<\/span>Dropout<span class=\"token punctuation\">(<\/span><span class=\"token number\">0.5<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p><span class=\"token comment\"># \u5728 forward \u4e2d&#xff0c;\u5168\u8fde\u63a5\u5c42\u4e4b\u95f4\u4f7f\u7528<\/span><br \/>\n<span class=\"token comment\"># x &#061; self.dropout(self.relu(self.fc1(x)))<\/span><\/p>\n<h4>8.4 \u6279\u5f52\u4e00\u5316<\/h4>\n<p>\u6279\u5f52\u4e00\u5316&#xff08;Batch Normalization&#xff09;\u5bf9\u6bcf\u4e00\u5c42\u7684\u8f93\u5165\u505a\u6807\u51c6\u5316&#xff0c;\u80fd\u52a0\u901f\u6536\u655b\u3001\u5141\u8bb8\u4f7f\u7528\u66f4\u5927\u7684\u5b66\u4e60\u7387&#xff0c;\u5e76\u8d77\u5230\u4e00\u5b9a\u7684\u6b63\u5219\u5316\u4f5c\u7528&#xff1a;<\/p>\n<p>self<span class=\"token punctuation\">.<\/span>bn1 <span class=\"token operator\">&#061;<\/span> nn<span class=\"token punctuation\">.<\/span>BatchNorm2d<span class=\"token punctuation\">(<\/span><span class=\"token number\">32<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p><span class=\"token comment\"># \u5728\u5377\u79ef\u5c42\u4e4b\u540e\u3001\u6fc0\u6d3b\u51fd\u6570\u4e4b\u524d\u4f7f\u7528<\/span><br \/>\n<span class=\"token comment\"># x &#061; self.relu(self.bn1(self.conv1(x)))<\/span><\/p>\n<h3>9. \u5e38\u89c1\u95ee\u9898\u4e0e\u8c03\u53c2\u5efa\u8bae<\/h3>\n<h4>9.1 \u6a21\u578b\u4e0d\u6536\u655b\u600e\u4e48\u529e<\/h4>\n<ul>\n<li>\u68c0\u67e5\u5b66\u4e60\u7387\u662f\u5426\u8fc7\u5927\u6216\u8fc7\u5c0f&#xff0c;\u53ef\u5c1d\u8bd5\u4ece 0.001 \u5f00\u59cb&#xff0c;\u7528\u5bf9\u6570\u5c3a\u5ea6\u641c\u7d22\u3002<\/li>\n<li>\u786e\u8ba4\u6570\u636e\u662f\u5426\u505a\u4e86\u5f52\u4e00\u5316&#xff0c;\u8f93\u5165\u8303\u56f4\u4e0d\u4e00\u81f4\u4f1a\u5bfc\u81f4\u68af\u5ea6\u9707\u8361\u3002<\/li>\n<li>\u68c0\u67e5\u635f\u5931\u51fd\u6570\u4e0e\u4efb\u52a1\u662f\u5426\u5339\u914d&#xff0c;\u5206\u7c7b\u4efb\u52a1\u7528\u4ea4\u53c9\u71b5&#xff0c;\u56de\u5f52\u4efb\u52a1\u7528 MSE\u3002<\/li>\n<\/ul>\n<h4>9.2 \u8fc7\u62df\u5408\u4e25\u91cd\u600e\u4e48\u529e<\/h4>\n<ul>\n<li>\u589e\u52a0\u6570\u636e\u589e\u5f3a&#xff0c;\u6269\u5145\u8bad\u7ec3\u6837\u672c\u591a\u6837\u6027\u3002<\/li>\n<li>\u5f15\u5165 Dropout \u6216 L2 \u6b63\u5219\u5316\u3002<\/li>\n<li>\u51cf\u5c0f\u6a21\u578b\u5bb9\u91cf&#xff0c;\u51cf\u5c11\u5377\u79ef\u6838\u6570\u91cf\u6216\u7f51\u7edc\u5c42\u6570\u3002<\/li>\n<li>\u4f7f\u7528\u65e9\u505c&#xff08;Early Stopping&#xff09;&#xff0c;\u5728\u9a8c\u8bc1\u96c6\u6307\u6807\u4e0d\u518d\u63d0\u5347\u65f6\u505c\u6b62\u8bad\u7ec3\u3002<\/li>\n<\/ul>\n<h4>9.3 \u663e\u5b58\u4e0d\u8db3\u600e\u4e48\u529e<\/h4>\n<ul>\n<li>\u51cf\u5c0f batch size&#xff0c;\u8fd9\u662f\u6700\u76f4\u63a5\u6709\u6548\u7684\u65b9\u6cd5\u3002<\/li>\n<li>\u964d\u4f4e\u8f93\u5165\u56fe\u50cf\u5206\u8fa8\u7387\u3002<\/li>\n<li>\u4f7f\u7528\u68af\u5ea6\u7d2f\u79ef&#xff0c;\u6a21\u62df\u66f4\u5927\u7684 batch size\u3002<\/li>\n<li>\u8003\u8651\u4f7f\u7528\u6df7\u5408\u7cbe\u5ea6\u8bad\u7ec3&#xff08;AMP&#xff09;&#xff0c;\u5728 PyTorch \u4e2d\u53ef\u901a\u8fc7 torch.cuda.amp \u5b9e\u73b0\u3002<\/li>\n<\/ul>\n<h3>10. \u8fdb\u9636\u5b66\u4e60\u8def\u7ebf<\/h3>\n<p>\u638c\u63e1\u57fa\u7840 CNN \u4e4b\u540e&#xff0c;\u53ef\u4ee5\u6cbf\u7740\u4ee5\u4e0b\u65b9\u5411\u7ee7\u7eed\u6df1\u5165&#xff1a;<\/p>\n<ul>\n<li>\u76ee\u6807\u68c0\u6d4b&#xff1a;\u5b66\u4e60 YOLO\u3001Faster R-CNN\u3001SSD \u7b49\u7ecf\u5178\u68c0\u6d4b\u6846\u67b6\u3002<\/li>\n<li>\u56fe\u50cf\u5206\u5272&#xff1a;\u638c\u63e1 U-Net\u3001Mask R-CNN \u7b49\u8bed\u4e49\/\u5b9e\u4f8b\u5206\u5272\u6a21\u578b\u3002<\/li>\n<li>\u8f7b\u91cf\u5316\u7f51\u7edc&#xff1a;\u4e86\u89e3 MobileNet\u3001ShuffleNet \u7b49\u9002\u5408\u79fb\u52a8\u7aef\u90e8\u7f72\u7684\u6a21\u578b\u3002<\/li>\n<li>\u6ce8\u610f\u529b\u673a\u5236&#xff1a;\u5b66\u4e60 SE-Net\u3001CBAM \u4ee5\u53ca Transformer \u5728\u89c6\u89c9\u4e2d\u7684\u5e94\u7528&#xff08;ViT\u3001Swin Transformer&#xff09;\u3002<\/li>\n<li>\u751f\u6210\u6a21\u578b&#xff1a;\u63a2\u7d22 GAN\u3001Diffusion Model \u5728\u56fe\u50cf\u751f\u6210\u9886\u57df\u7684\u5e94\u7528\u3002<\/li>\n<\/ul>\n<h3>11. \u5e38\u89c1\u95ee\u9898 FAQ<\/h3>\n<p>Q1&#xff1a;CNN \u53ea\u80fd\u5904\u7406\u56fe\u50cf\u5417&#xff1f;<\/p>\n<p>\u4e0d\u662f\u3002CNN \u9002\u7528\u4e8e\u4efb\u4f55\u5177\u6709\u7f51\u683c\u7ed3\u6784\u7684\u6570\u636e&#xff0c;\u5305\u62ec\u97f3\u9891&#xff08;\u4e00\u7ef4\u7f51\u683c&#xff09;\u3001\u89c6\u9891&#xff08;\u4e09\u7ef4\u7f51\u683c&#xff09;\u4ee5\u53ca\u90e8\u5206\u5e8f\u5217\u6570\u636e\u3002\u5728\u81ea\u7136\u8bed\u8a00\u5904\u7406\u4e2d&#xff0c;CNN \u4e5f\u66fe\u88ab\u7528\u4e8e\u6587\u672c\u5206\u7c7b\u7b49\u4efb\u52a1\u3002<\/p>\n<p>Q2&#xff1a;\u5377\u79ef\u6838\u5927\u5c0f\u5982\u4f55\u9009\u62e9&#xff1f;<\/p>\n<p>\u5c0f\u5377\u79ef\u6838&#xff08;\u5982 3\u00d73&#xff09;\u53c2\u6570\u91cf\u5c11\u3001\u53ef\u4ee5\u5806\u53e0\u66f4\u6df1&#xff0c;\u662f\u5f53\u524d\u4e3b\u6d41\u9009\u62e9&#xff1b;\u5927\u5377\u79ef\u6838&#xff08;\u5982 7\u00d77&#xff09;\u611f\u53d7\u91ce\u66f4\u5927&#xff0c;\u4f46\u53c2\u6570\u91cf\u548c\u8ba1\u7b97\u91cf\u4e5f\u66f4\u5927\u3002VGG \u8bc1\u660e\u4e86\u591a\u4e2a\u5c0f\u5377\u79ef\u6838\u5806\u53e0\u53ef\u4ee5\u7b49\u6548\u4e8e\u5927\u5377\u79ef\u6838\u7684\u611f\u53d7\u91ce\u3002<\/p>\n<p>Q3&#xff1a;\u6c60\u5316\u5c42\u53ef\u4ee5\u53bb\u6389\u5417&#xff1f;<\/p>\n<p>\u53ef\u4ee5&#xff0c;\u4f46\u901a\u5e38\u4e0d\u5efa\u8bae\u3002\u6c60\u5316\u5c42\u80fd\u964d\u4f4e\u7279\u5f81\u56fe\u5206\u8fa8\u7387\u3001\u51cf\u5c11\u8ba1\u7b97\u91cf\u5e76\u589e\u5f3a\u5e73\u79fb\u4e0d\u53d8\u6027\u3002\u5982\u679c\u53bb\u6389\u6c60\u5316&#xff0c;\u53ef\u4ee5\u901a\u8fc7\u589e\u5927\u5377\u79ef\u6b65\u957f&#xff08;stride&#xff09;\u6765\u66ff\u4ee3\u4e0b\u91c7\u6837\u3002<\/p>\n<p>Q4&#xff1a;\u4e3a\u4ec0\u4e48 CNN \u6bd4\u5168\u8fde\u63a5\u7f51\u7edc\u66f4\u9002\u5408\u56fe\u50cf&#xff1f;<\/p>\n<p>\u56e0\u4e3a CNN \u5229\u7528\u4e86\u56fe\u50cf\u7684\u5c40\u90e8\u76f8\u5173\u6027\u548c\u5e73\u79fb\u4e0d\u53d8\u6027&#xff0c;\u901a\u8fc7\u6743\u503c\u5171\u4eab\u5927\u5e45\u51cf\u5c11\u53c2\u6570\u91cf&#xff0c;\u540c\u65f6\u4fdd\u7559\u4e86\u7a7a\u95f4\u7ed3\u6784\u4fe1\u606f&#xff0c;\u56e0\u6b64\u5728\u5c0f\u6837\u672c\u4e0b\u4e5f\u80fd\u53d6\u5f97\u66f4\u597d\u7684\u6cdb\u5316\u6548\u679c\u3002<\/p>\n<p>CNN \u901a\u8fc7\u5377\u79ef\u3001\u6c60\u5316\u548c\u5168\u8fde\u63a5\u5c42\u7684\u7ec4\u5408&#xff0c;\u4ee5\u5c40\u90e8\u8fde\u63a5\u548c\u6743\u503c\u5171\u4eab\u7684\u65b9\u5f0f\u9ad8\u6548\u63d0\u53d6\u56fe\u50cf\u7279\u5f81&#xff0c;\u662f\u8ba1\u7b97\u673a\u89c6\u89c9\u9886\u57df\u7684\u57fa\u77f3\u3002\u672c\u6587\u4ece\u539f\u7406\u5230\u4ee3\u7801\u5b9e\u6218&#xff0c;\u5e26\u4f60\u5b8c\u6574\u8d70\u901a\u4e86 CNN \u7684\u6838\u5fc3\u6d41\u7a0b\u3002\u638c\u63e1 CNN \u540e&#xff0c;\u53ef\u4ee5\u8fdb\u4e00\u6b65\u5b66\u4e60\u76ee\u6807\u68c0\u6d4b&#xff08;YOLO\u3001Faster R-CNN&#xff09;\u3001\u56fe\u50cf\u5206\u5272&#xff08;U-Net\u3001Mask R-CNN&#xff09;\u7b49\u66f4\u590d\u6742\u7684\u89c6\u89c9\u4efb\u52a1\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"<p>1. \u5f15\u8a00<br \/>\n\u5377\u79ef\u795e\u7ecf\u7f51\u7edc&#xff08;Convolutional Neural Network&#xff0c;\u7b80\u79f0 CNN&#xff09;\u662f\u6df1\u5ea6\u5b66\u4e60\u9886\u57df\u6700\u5177\u5f71\u54cd\u529b\u7684\u6a21\u578b\u4e4b\u4e00&#xff0c;\u5c24\u5176\u5728\u8ba1\u7b97\u673a\u89c6\u89c9\u4efb\u52a1\u4e2d\u8868\u73b0\u5353\u8d8a\u3002\u4ece 2012 \u5e74 AlexNet \u5728 ImageNet \u7ade\u8d5b\u4e2d\u4e00\u4e3e\u593a\u51a0\u5f00\u59cb&#xff0c;CNN \u4fbf\u6210\u4e3a\u56fe\u50cf\u8bc6\u522b\u3001\u76ee\u6807\u68c0\u6d4b\u3001\u56fe\u50cf\u5206\u5272\u7b49\u4efb\u52a1\u7684\u4e3b\u6d41\u65b9\u6848\u3002\u672c\u6587\u5c06\u5e26\u4f60\u4ece\u96f6\u7406\u89e3 CNN \u7684\u6838\u5fc3\u539f\u7406&#xff0c;\u5e76\u901a\u8fc7\u4ee3\u7801\u5b9e\u6218\u638c\u63e1\u5176\u5e94\u7528\u65b9\u6cd5\u3002<br \/>\n2. \u4ec0\u4e48\u662f\u5377\u79ef\u795e\u7ecf\u7f51<\/p>\n","protected":false},"author":2,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[587,2394,50],"topic":[],"class_list":["post-114911","post","type-post","status-publish","format-standard","hentry","category-server","tag-587","tag-cnn","tag-50"],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v20.3 - https:\/\/yoast.com\/wordpress\/plugins\/seo\/ -->\n<title>CNN \u5377\u79ef\u795e\u7ecf\u7f51\u7edc\u8be6\u89e3\uff1a\u4ece\u539f\u7406\u5230\u5b9e\u6218\uff082\uff09 - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/www.wsisp.com\/helps\/114911.html\" \/>\n<meta property=\"og:locale\" content=\"zh_CN\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"CNN \u5377\u79ef\u795e\u7ecf\u7f51\u7edc\u8be6\u89e3\uff1a\u4ece\u539f\u7406\u5230\u5b9e\u6218\uff082\uff09 - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\" \/>\n<meta property=\"og:description\" content=\"1. \u5f15\u8a00 \u5377\u79ef\u795e\u7ecf\u7f51\u7edc&#xff08;Convolutional Neural Network&#xff0c;\u7b80\u79f0 CNN&#xff09;\u662f\u6df1\u5ea6\u5b66\u4e60\u9886\u57df\u6700\u5177\u5f71\u54cd\u529b\u7684\u6a21\u578b\u4e4b\u4e00&#xff0c;\u5c24\u5176\u5728\u8ba1\u7b97\u673a\u89c6\u89c9\u4efb\u52a1\u4e2d\u8868\u73b0\u5353\u8d8a\u3002\u4ece 2012 \u5e74 AlexNet \u5728 ImageNet \u7ade\u8d5b\u4e2d\u4e00\u4e3e\u593a\u51a0\u5f00\u59cb&#xff0c;CNN \u4fbf\u6210\u4e3a\u56fe\u50cf\u8bc6\u522b\u3001\u76ee\u6807\u68c0\u6d4b\u3001\u56fe\u50cf\u5206\u5272\u7b49\u4efb\u52a1\u7684\u4e3b\u6d41\u65b9\u6848\u3002\u672c\u6587\u5c06\u5e26\u4f60\u4ece\u96f6\u7406\u89e3 CNN \u7684\u6838\u5fc3\u539f\u7406&#xff0c;\u5e76\u901a\u8fc7\u4ee3\u7801\u5b9e\u6218\u638c\u63e1\u5176\u5e94\u7528\u65b9\u6cd5\u3002 2. \u4ec0\u4e48\u662f\u5377\u79ef\u795e\u7ecf\u7f51\" \/>\n<meta property=\"og:url\" content=\"https:\/\/www.wsisp.com\/helps\/114911.html\" \/>\n<meta property=\"og:site_name\" content=\"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\" \/>\n<meta property=\"article:published_time\" content=\"2026-10-10T05:57:04+00:00\" \/>\n<meta name=\"author\" content=\"admin\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"\u4f5c\u8005\" \/>\n\t<meta name=\"twitter:data1\" content=\"admin\" \/>\n\t<meta name=\"twitter:label2\" content=\"\u9884\u8ba1\u9605\u8bfb\u65f6\u95f4\" \/>\n\t<meta name=\"twitter:data2\" content=\"7 \u5206\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/114911.html\",\"url\":\"https:\/\/www.wsisp.com\/helps\/114911.html\",\"name\":\"CNN \u5377\u79ef\u795e\u7ecf\u7f51\u7edc\u8be6\u89e3\uff1a\u4ece\u539f\u7406\u5230\u5b9e\u6218\uff082\uff09 - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\",\"isPartOf\":{\"@id\":\"https:\/\/www.wsisp.com\/helps\/#website\"},\"datePublished\":\"2026-10-10T05:57:04+00:00\",\"dateModified\":\"2026-10-10T05:57:04+00:00\",\"author\":{\"@id\":\"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41\"},\"breadcrumb\":{\"@id\":\"https:\/\/www.wsisp.com\/helps\/114911.html#breadcrumb\"},\"inLanguage\":\"zh-Hans\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/www.wsisp.com\/helps\/114911.html\"]}]},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/114911.html#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"\u9996\u9875\",\"item\":\"https:\/\/www.wsisp.com\/helps\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"CNN \u5377\u79ef\u795e\u7ecf\u7f51\u7edc\u8be6\u89e3\uff1a\u4ece\u539f\u7406\u5230\u5b9e\u6218\uff082\uff09\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/#website\",\"url\":\"https:\/\/www.wsisp.com\/helps\/\",\"name\":\"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\",\"description\":\"\u9999\u6e2f\u670d\u52a1\u5668_\u9999\u6e2f\u4e91\u670d\u52a1\u5668\u8d44\u8baf_\u670d\u52a1\u5668\u5e2e\u52a9\u6587\u6863_\u670d\u52a1\u5668\u6559\u7a0b\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/www.wsisp.com\/helps\/?s={search_term_string}\"},\"query-input\":\"required name=search_term_string\"}],\"inLanguage\":\"zh-Hans\"},{\"@type\":\"Person\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41\",\"name\":\"admin\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"zh-Hans\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/image\/\",\"url\":\"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery\",\"contentUrl\":\"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery\",\"caption\":\"admin\"},\"sameAs\":[\"http:\/\/wp.wsisp.com\"],\"url\":\"https:\/\/www.wsisp.com\/helps\/author\/admin\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"CNN \u5377\u79ef\u795e\u7ecf\u7f51\u7edc\u8be6\u89e3\uff1a\u4ece\u539f\u7406\u5230\u5b9e\u6218\uff082\uff09 - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/www.wsisp.com\/helps\/114911.html","og_locale":"zh_CN","og_type":"article","og_title":"CNN \u5377\u79ef\u795e\u7ecf\u7f51\u7edc\u8be6\u89e3\uff1a\u4ece\u539f\u7406\u5230\u5b9e\u6218\uff082\uff09 - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","og_description":"1. \u5f15\u8a00 \u5377\u79ef\u795e\u7ecf\u7f51\u7edc&#xff08;Convolutional Neural Network&#xff0c;\u7b80\u79f0 CNN&#xff09;\u662f\u6df1\u5ea6\u5b66\u4e60\u9886\u57df\u6700\u5177\u5f71\u54cd\u529b\u7684\u6a21\u578b\u4e4b\u4e00&#xff0c;\u5c24\u5176\u5728\u8ba1\u7b97\u673a\u89c6\u89c9\u4efb\u52a1\u4e2d\u8868\u73b0\u5353\u8d8a\u3002\u4ece 2012 \u5e74 AlexNet \u5728 ImageNet \u7ade\u8d5b\u4e2d\u4e00\u4e3e\u593a\u51a0\u5f00\u59cb&#xff0c;CNN \u4fbf\u6210\u4e3a\u56fe\u50cf\u8bc6\u522b\u3001\u76ee\u6807\u68c0\u6d4b\u3001\u56fe\u50cf\u5206\u5272\u7b49\u4efb\u52a1\u7684\u4e3b\u6d41\u65b9\u6848\u3002\u672c\u6587\u5c06\u5e26\u4f60\u4ece\u96f6\u7406\u89e3 CNN \u7684\u6838\u5fc3\u539f\u7406&#xff0c;\u5e76\u901a\u8fc7\u4ee3\u7801\u5b9e\u6218\u638c\u63e1\u5176\u5e94\u7528\u65b9\u6cd5\u3002 2. \u4ec0\u4e48\u662f\u5377\u79ef\u795e\u7ecf\u7f51","og_url":"https:\/\/www.wsisp.com\/helps\/114911.html","og_site_name":"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","article_published_time":"2026-10-10T05:57:04+00:00","author":"admin","twitter_card":"summary_large_image","twitter_misc":{"\u4f5c\u8005":"admin","\u9884\u8ba1\u9605\u8bfb\u65f6\u95f4":"7 \u5206"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/www.wsisp.com\/helps\/114911.html","url":"https:\/\/www.wsisp.com\/helps\/114911.html","name":"CNN \u5377\u79ef\u795e\u7ecf\u7f51\u7edc\u8be6\u89e3\uff1a\u4ece\u539f\u7406\u5230\u5b9e\u6218\uff082\uff09 - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","isPartOf":{"@id":"https:\/\/www.wsisp.com\/helps\/#website"},"datePublished":"2026-10-10T05:57:04+00:00","dateModified":"2026-10-10T05:57:04+00:00","author":{"@id":"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41"},"breadcrumb":{"@id":"https:\/\/www.wsisp.com\/helps\/114911.html#breadcrumb"},"inLanguage":"zh-Hans","potentialAction":[{"@type":"ReadAction","target":["https:\/\/www.wsisp.com\/helps\/114911.html"]}]},{"@type":"BreadcrumbList","@id":"https:\/\/www.wsisp.com\/helps\/114911.html#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"\u9996\u9875","item":"https:\/\/www.wsisp.com\/helps"},{"@type":"ListItem","position":2,"name":"CNN \u5377\u79ef\u795e\u7ecf\u7f51\u7edc\u8be6\u89e3\uff1a\u4ece\u539f\u7406\u5230\u5b9e\u6218\uff082\uff09"}]},{"@type":"WebSite","@id":"https:\/\/www.wsisp.com\/helps\/#website","url":"https:\/\/www.wsisp.com\/helps\/","name":"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","description":"\u9999\u6e2f\u670d\u52a1\u5668_\u9999\u6e2f\u4e91\u670d\u52a1\u5668\u8d44\u8baf_\u670d\u52a1\u5668\u5e2e\u52a9\u6587\u6863_\u670d\u52a1\u5668\u6559\u7a0b","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/www.wsisp.com\/helps\/?s={search_term_string}"},"query-input":"required name=search_term_string"}],"inLanguage":"zh-Hans"},{"@type":"Person","@id":"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41","name":"admin","image":{"@type":"ImageObject","inLanguage":"zh-Hans","@id":"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/image\/","url":"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery","contentUrl":"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery","caption":"admin"},"sameAs":["http:\/\/wp.wsisp.com"],"url":"https:\/\/www.wsisp.com\/helps\/author\/admin"}]}},"_links":{"self":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/posts\/114911","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/comments?post=114911"}],"version-history":[{"count":0,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/posts\/114911\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/media?parent=114911"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/categories?post=114911"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/tags?post=114911"},{"taxonomy":"topic","embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/topic?post=114911"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}