{"id":3824,"date":"2025-11-06T19:49:57","date_gmt":"2025-11-06T11:49:57","guid":{"rendered":"https:\/\/www.leexinghai.com\/aic\/?p=3824"},"modified":"2025-11-24T18:42:33","modified_gmt":"2025-11-24T10:42:33","slug":"%e8%ae%ba%e6%96%87%e8%af%84%e8%bf%b0-%e6%96%87%e7%8c%aefrt-fs-en-2510282","status":"publish","type":"post","link":"https:\/\/www.leexinghai.com\/aic\/%e8%ae%ba%e6%96%87%e8%af%84%e8%bf%b0-%e6%96%87%e7%8c%aefrt-fs-en-2510282\/","title":{"rendered":"\u8bba\u6587\u8bc4\u8ff0-\u6587\u732eFRT-FS-EN-2510282"},"content":{"rendered":"\n<p>\u672c\u671f\u8bc4\u8ff0\u6587\u7ae0\uff1a<\/p>\n\n\n\n<figure class=\"wp-block-embed is-type-wp-embed\"><div class=\"wp-block-embed__wrapper\">\n<blockquote class=\"wp-embedded-content\" data-secret=\"eYbUY0AYsg\"><a href=\"https:\/\/www.leexinghai.com\/aic\/frt-fs-en-2510282\/\">[\u6587\u732eFRT-FS-EN-2510282]Few-shot crop disease recognition using sequence- weighted ensemble model-agnostic meta-learning<\/a><\/blockquote><iframe loading=\"lazy\" class=\"wp-embedded-content\" sandbox=\"allow-scripts\" security=\"restricted\" style=\"position: absolute; visibility: hidden;\" title=\"\u300a [\u6587\u732eFRT-FS-EN-2510282]Few-shot crop disease recognition using sequence- weighted ensemble model-agnostic meta-learning \u300b\u2014\u5b66\u672f\u521b\u65b0\u4e2d\u5fc3\" src=\"https:\/\/www.leexinghai.com\/aic\/frt-fs-en-2510282\/embed\/#?secret=YHEEJQJ3Xo#?secret=eYbUY0AYsg\" data-secret=\"eYbUY0AYsg\" width=\"600\" height=\"338\" frameborder=\"0\" marginwidth=\"0\" marginheight=\"0\" scrolling=\"no\"><\/iframe>\n<\/div><\/figure>\n\n\n\n<p>\u5b83\u63a2\u8ba8\u7684\u662f\u4e00\u4e2a\u5728\u519c\u4e1a\u548cAI\u4ea4\u53c9\u9886\u57df\u975e\u5e38\u91cd\u8981\u7684\u95ee\u9898\uff1a\u5982\u4f55\u5728\u6570\u636e\u6837\u672c\u5f88\u5c11\u7684\u60c5\u51b5\u4e0b\uff08\u5373\u201c\u5c0f\u6837\u672c\u5b66\u4e60\u201d\uff09\u51c6\u786e\u8bc6\u522b\u4f5c\u7269\u75c5\u5bb3 \u3002<\/p>\n\n\n\n<p>\u4e3a\u4e86\u89e3\u51b3\u8fd9\u4e2a\u95ee\u9898\uff0c\u8bba\u6587\u63d0\u51fa\u4e86\u4e00\u79cd\u540d\u4e3a <strong>SWE-MAML<\/strong>\uff08\u5e8f\u5217\u52a0\u6743\u96c6\u6210\u6a21\u578b\u65e0\u5173\u5143\u5b66\u4e60\uff09\u7684\u65b0\u65b9\u6cd5 \u3002<\/p>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity is-style-dots\"\/>\n\n\n\n<h2 class=\"wp-block-heading has-text-align-center\">1.\u4e3a\u4ec0\u4e48WHY<\/h2>\n\n\n\n<blockquote class=\"wp-block-quote is-layout-flow wp-block-quote-is-layout-flow\">\n<p>While deep learning-based computer vision techniques have emerged as powerful tools for crop disease recognition, these methods are heavily reliant on large datasets, which are often difficult to obtain in practical agricultural settings. <\/p>\n\n\n\n<p>Trans:\u867d\u7136\u57fa\u4e8e\u6df1\u5ea6\u5b66\u4e60\u7684\u8ba1\u7b97\u673a\u89c6\u89c9\u6280\u672f\u5df2\u6210\u4e3a\u519c\u4f5c\u7269\u75c5\u5bb3\u8bc6\u522b\u7684\u5229\u5668\uff0c\u4f46\u8fd9\u4e9b\u65b9\u6cd5\u4e25\u91cd\u4f9d\u8d56\u6d77\u91cf\u6570\u636e\u96c6\uff0c\u800c\u5b9e\u9645\u519c\u4e1a\u573a\u666f\u4e2d\u5f80\u5f80\u96be\u4ee5\u83b7\u53d6\u3002<\/p>\n<\/blockquote>\n\n\n\n<p>\u8fd9\u7bc7\u8bba\u6587\u7684\u6838\u5fc3\u662f\u89e3\u51b3\u4e00\u4e2a\u5728\u73b0\u5b9e\u4e2d\u975e\u5e38\u68d8\u624b\u7684\u95ee\u9898\u3002\u4f20\u7edf\u7684\u6df1\u5ea6\u5b66\u4e60\u6a21\u578b\uff08\u6bd4\u5982CNN\uff09\u5728\u56fe\u50cf\u8bc6\u522b\u4e0a\u6548\u679c\u5f88\u597d\uff0c\u4f46\u5b83\u4eec\u901a\u5e38\u9700\u8981\u4e00\u4e2a\u524d\u63d0\u6761\u4ef6\uff1a\u6d77\u91cf\u7684\u8bad\u7ec3\u6570\u636e \u3002<\/p>\n\n\n\n<p>\u4f46\u5728\u519c\u4e1a\u9886\u57df\uff0c\u8981\u83b7\u53d6\u5927\u91cf\u3001\u591a\u6837\u5316\u7684\u75c5\u5bb3\u56fe\u50cf\uff0c\u6070\u6070\u662f\u975e\u5e38\u56f0\u96be\u7684 \u3002<\/p>\n\n\n\n<p>\u3010\u5c0f\u95ee1\u3011\u90a3\u4e48\uff0c\u6839\u636e\u8bba\u6587\u7684\u5f15\u8a00\uff08Introduction\uff09\u90e8\u5206\uff0c\u4f60\u8ba4\u4e3a\u5177\u4f53\u662f\u4ec0\u4e48\u56e0\u7d20\u5bfc\u81f4\u4e86\u6536\u96c6\u5927\u91cf\u4f5c\u7269\u75c5\u5bb3\u6570\u636e\u5982\u6b64\u56f0\u96be\u6216\u6602\u8d35\u5462\uff1f<\/p>\n\n\n\n<blockquote class=\"wp-block-quote is-layout-flow wp-block-quote-is-layout-flow\">\n<p>\u3010\u56de\u7b541\u3011<\/p>\n\n\n\n<p>Agricultural scenarios are usually complex and diverse. Crop disease occurrence has a long time and space span in general.<\/p>\n\n\n\n<p>Trans:\u519c\u4e1a\u60c5\u666f\u901a\u5e38\u590d\u6742\u591a\u6837\u3002\u4f5c\u7269\u75c5\u5bb3\u7684\u53d1\u751f\u901a\u5e38\u5177\u6709\u957f\u671f\u6027\u548c\u7a7a\u95f4\u5206\u5e03\u7279\u5f81\u3002<\/p>\n\n\n\n<p>Furthermore, crop diseases spread widely in time and space, and the annotation of disease data needs to be done manually by experienced experts. Therefore, large-scale disease image collection and annotation is very costly.<\/p>\n\n\n\n<p>Trans:\u6b64\u5916\uff0c\u519c\u4f5c\u7269\u75c5\u5bb3\u5728\u65f6\u7a7a\u4e0a\u5e7f\u6cdb\u4f20\u64ad\uff0c\u4e14\u75c5\u5bb3\u6570\u636e\u6807\u6ce8\u9700\u7531\u7ecf\u9a8c\u4e30\u5bcc\u7684\u4e13\u5bb6\u4eba\u5de5\u5b8c\u6210\u3002\u56e0\u6b64\uff0c\u5927\u89c4\u6a21\u75c5\u5bb3\u56fe\u50cf\u91c7\u96c6\u4e0e\u6807\u6ce8\u6210\u672c\u6781\u9ad8\u3002<\/p>\n<\/blockquote>\n\n\n\n<p>\u603b\u7ed3\u4e86\u201c\u4e3a\u4ec0\u4e48\u201d\u8fd9\u4e2a\u95ee\u9898\uff1a<\/p>\n\n\n\n<ol class=\"wp-block-list\">\n<li>\u73af\u5883\u590d\u6742\uff1a\u75c5\u5bb3\u5728\u4e0d\u540c\u751f\u957f\u671f\u548c\u73af\u5883\u4e0b\u8868\u73b0\u4e0d\u540c\u3002<\/li>\n\n\n\n<li>\u6807\u6ce8\u6602\u8d35\uff1a\u9700\u8981\u7ecf\u9a8c\u4e30\u5bcc\u7684\u4e13\u5bb6\u6765\u624b\u52a8\u6807\u6ce8\uff0c\u6210\u672c\u5f88\u9ad8\u3002<\/li>\n<\/ol>\n\n\n\n<p>\u8fd9\u5c31\u4ea7\u751f\u4e86\u4e00\u4e2a\u6838\u5fc3\u77db\u76fe\uff1a\u6df1\u5ea6\u5b66\u4e60\u6a21\u578b\u6e34\u671b\u201c\u5927\u6570\u636e\u201d\uff0c\u4f46\u73b0\u5b9e\u519c\u4e1a\u573a\u666f\u5374\u96be\u4ee5\u63d0\u4f9b\u3002<\/p>\n\n\n\n<p>\u4e3a\u4e86\u89e3\u51b3\u8fd9\u4e2a\u201c\u6570\u636e\u7a00\u7f3a\u201d\u7684\u77db\u76fe\uff0c\u8bba\u6587\uff08\u4ee5\u53ca\u6574\u4e2a\u7814\u7a76\u9886\u57df\uff09\u63d0\u51fa\u4e86\u4e00\u79cd\u4e13\u95e8\u7684\u89e3\u51b3\u65b9\u6848\u3002\u8fd9\u7bc7\u8bba\u6587\uff08\u7b2c2\u9875\uff09\u5c06\u5176\u79f0\u4e3a<strong>\u201c\u5c0f\u6837\u672c\u5b66\u4e60\u201d\uff08Few-shot learning, FSL\uff09<\/strong>\u3002<\/p>\n\n\n\n<p>\u3010\u5c0f\u95ee2\u3011\u6839\u636e\u8bba\u6587\u5bf9FSL\u7684\u63cf\u8ff0\uff0c\u4f60\u8ba4\u4e3a\u201c\u5c0f\u6837\u672c\u5b66\u4e60\u201d\u7684\u76ee\u6807\u662f\u4ec0\u4e48\uff1f\u5b83\u4e0e\u4f7f\u7528\u6d77\u91cf\u6570\u636e\u7684\u201c\u6807\u51c6\u201d\u6df1\u5ea6\u5b66\u4e60\u6709\u4f55\u4e0d\u540c\uff1f<\/p>\n\n\n\n<blockquote class=\"wp-block-quote is-layout-flow wp-block-quote-is-layout-flow\">\n<p>\u3010\u56de\u7b542\u3011As the name implies, few-shot learning refers to feeding a learning model with a tiny amount of training data, contrary to the standard practice of using a large amount of data.<\/p>\n\n\n\n<p>Trans:\u987e\u540d\u601d\u4e49\uff0c\u5c0f\u6837\u672c\u5b66\u4e60\u662f\u6307\u7528\u5c11\u91cf\u8bad\u7ec3\u6570\u636e\u8bad\u7ec3\u6a21\u578b\uff0c\u8fd9\u4e0e\u5e38\u89c4\u4f7f\u7528\u5927\u91cf\u6570\u636e\u7684\u505a\u6cd5\u622a\u7136\u4e0d\u540c\u3002<\/p>\n<\/blockquote>\n\n\n\n<p>\u73b0\u5728\u6211\u4eec\u77e5\u9053\u4e86\u201c\u4e3a\u4ec0\u4e48\u201d\u9700\u8981FSL\uff08\u56e0\u4e3a\u6536\u96c6\u6570\u636e\u56f0\u96be\uff09\uff0c\u4ee5\u53caFSL\u7684\u201c\u76ee\u6807\u201d\uff08\u7528\u5c11\u91cf\u6570\u636e\u8bad\u7ec3\uff09\u3002<\/p>\n\n\n\n<p>\u3010\u5c0f\u95ee3\u3011\u5728\u5f15\u8a00\uff08Introduction\uff09\u7684\u540e\u534a\u90e8\u5206\uff08\u7b2c2\u9875\uff09\uff0c\u8bba\u6587\u63d0\u5230\u4e86\u51e0\u79cd\u5b9e\u73b0FSL\u7684\u4e3b\u8981\u6280\u672f\u9014\u5f84\u3002\u4f60\u80fd\u627e\u5230\u5b83\u4eec\u5417\uff1f\u8bba\u6587\u5c06FSL\u65b9\u6cd5\u4e3b\u8981\u5206\u4e3a\u4e86\u54ea\u4e09\u7c7b\uff1f<\/p>\n\n\n\n<blockquote class=\"wp-block-quote is-layout-flow wp-block-quote-is-layout-flow\">\n<p>\u3010\u56de\u7b543\u3011Few-shot learning for image recognition can be mainly grouped into model initialization, metric learning, and data augmentation.<\/p>\n\n\n\n<p>\u56fe\u50cf\u8bc6\u522b\u9886\u57df\u7684\u5c11\u6837\u672c\u5b66\u4e60\u6280\u672f\u4e3b\u8981\u5305\u542b\u4e09\u5927\u65b9\u5411\uff1a\u6a21\u578b\u521d\u59cb\u5316\u3001\u5ea6\u91cf\u5b66\u4e60\u548c\u6570\u636e\u589e\u5f3a\u3002<\/p>\n<\/blockquote>\n\n\n\n<p>\u73b0\u5728\uff0c\u5f15\u8a00\u90e8\u5206\uff08\u9875\u9762\u7b2c 2 \u548c 3 \u90e8\u5206\uff09\u5c06\u4ece\u8fd9\u4e9b\u603b\u4f53\u6982\u5ff5\u8fc7\u6e21\u5230\u8be5\u9886\u57df\u975e\u5e38\u6d41\u884c\u7684\u4e00\u79cd\u7279\u5b9a\u7b97\u6cd5\uff0c\u8be5\u7b97\u6cd5\u4e5f\u662f\u672c\u6587\u65b0\u65b9\u6cd5\u7684\u57fa\u7840\u3002<\/p>\n\n\n\n<p>\u3010\u5c0f\u95ee4\u3011\u4f60\u80fd\u627e\u5230\u6587\u4e2d\u63d0\u5230\u7684\u8fd9\u79cd\u6d41\u884c\u7684 FSL \u7b97\u6cd5\u7684\u540d\u79f0\u5417\uff1f\u5b83\u5c5e\u4e8e\u8fd9\u4e09\u5927\u7c7b\u522b\u4e2d\u7684\u54ea\u4e00\u7c7b\uff08\u6a21\u578b\u521d\u59cb\u5316\u3001\u5ea6\u91cf\u5b66\u4e60\u6216\u6570\u636e\u589e\u5f3a\uff09\uff1f<\/p>\n\n\n\n<blockquote class=\"wp-block-quote is-layout-flow wp-block-quote-is-layout-flow\">\n<p>\u3010\u56de\u7b544\u3011Model-Agnostic Meta-Learning (MAML) is one of the most popular few-shot learning algorithms (Finn et al., 2017).<\/p>\n\n\n\n<p>Trans:\u6a21\u578b\u65e0\u5173\u5143\u5b66\u4e60\uff08MAML\uff09\u662f\u5f53\u524d\u6700\u4e3b\u6d41\u7684\u5c11\u6837\u672c\u5b66\u4e60\u7b97\u6cd5\u4e4b\u4e00\uff08Finn\u7b49\uff0c2017\uff09\u3002<\/p>\n\n\n\n<p>MAML is a meta-learning framework based on <em><strong>model initialization<\/strong><\/em> by training the model\u2019s parameters so that a small number of gradient updates are going to lead to fast learning on a novel task.<\/p>\n\n\n\n<p>Trans:MAML\u662f\u4e00\u79cd\u5143\u5b66\u4e60\u6846\u67b6\uff0c\u5176\u6838\u5fc3\u5728\u4e8e\u901a\u8fc7<strong>\u8bad\u7ec3\u6a21\u578b\u53c2\u6570\u8fdb\u884c\u521d\u59cb\u5316<\/strong>\uff0c\u4f7f\u5f97\u5c11\u91cf\u68af\u5ea6\u66f4\u65b0\u5373\u53ef\u5728\u65b0\u4efb\u52a1\u4e0a\u5b9e\u73b0\u5feb\u901f\u5b66\u4e60\u3002<\/p>\n<\/blockquote>\n\n\n\n<p>\u5b83\u7684\u6838\u5fc3\u601d\u60f3\u4e0d\u662f\u8bad\u7ec3\u4e00\u4e2a\u201c\u6700\u7ec8\u6a21\u578b\u201d\uff0c\u800c\u662f\u8bad\u7ec3\u4e00\u4e2a\u201c\u521d\u59cb\u6a21\u578b\u201d\u3002<\/p>\n\n\n\n<p>\u90a3\u4e48\uff0c\u6839\u636e\u8bba\u6587\u7b2c3\u9875\u5bf9 MAML \u7684\u63cf\u8ff0 \uff0cMAML \u8bad\u7ec3\u51fa\u7684\u8fd9\u5957\u201c\u521d\u59cb\u5316\u53c2\u6570\u201d\u6709\u4ec0\u4e48\u7279\u522b\u4e4b\u5904\uff1f\u5b83\u80fd\u8ba9\u6a21\u578b\u5728\u9047\u5230\u4e00\u4e2a\u65b0\u4efb\u52a1\u65f6\u5b9e\u73b0\u4ec0\u4e48\u6548\u679c\uff1f<\/p>\n\n\n\n<blockquote class=\"wp-block-quote is-layout-flow wp-block-quote-is-layout-flow\">\n<p>MAML is a meta-learning framework based on model initialization by training the model\u2019s parameters so that a small number of gradient updates are going to lead to fast learning on a novel task. <\/p>\n\n\n\n<p>Trans:MAML\u662f\u4e00\u79cd\u5143\u5b66\u4e60\u6846\u67b6\uff0c\u5176\u6838\u5fc3\u5728\u4e8e\u901a\u8fc7\u8bad\u7ec3\u6a21\u578b\u53c2\u6570\u8fdb\u884c\u521d\u59cb\u5316\uff0c\u4f7f\u5f97\u5c11\u91cf\u68af\u5ea6\u66f4\u65b0\u5c31\u80fd\u5728\u65b0\u4efb\u52a1\u4e0a\u5b9e\u73b0\u5feb\u901f\u5b66\u4e60\u3002<\/p>\n<\/blockquote>\n\n\n\n<p>MAML \u7684\u76ee\u6807\u5c31\u662f\u627e\u5230\u4e00\u5957\u201c\u5143\u201d\u53c2\u6570 \u03b8\uff0c\u5f53\u9762\u5bf9\u4e00\u4e2a\u65b0\u4efb\u52a1\u65f6\uff0c\u6a21\u578b\u53ea\u9700\u8981\u5728\u8fd9\u5957\u53c2\u6570\u7684\u57fa\u7840\u4e0a\u201c\u5fae\u8c03\u201d\u51e0\u6b65\uff08a small number of gradient updates\uff09\uff0c\u5c31\u80fd\u7acb\u523b\u9002\u5e94\u8fd9\u4e2a\u65b0\u4efb\u52a1\u5e76\u8868\u73b0\u826f\u597d\u3002 \u8fd9\u5c31\u662f\u201c\u5feb\u901f\u5b66\u4e60\u201d\u7684\u542b\u4e49\u3002<\/p>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<p>\u6211\u4eec\u73b0\u5728\u77e5\u9053\u4e86\u201c\u4e3a\u4ec0\u4e48\u201d\u9700\u8981 FSL\uff0c\u4e5f\u77e5\u9053\u4e86 MAML \u662f\u4e00\u4e2a\u57fa\u4e8e\u201c\u6a21\u578b\u521d\u59cb\u5316\u201d\u7684\u5de7\u5999\u65b9\u6848\u3002<\/p>\n\n\n\n<p>\u4f46\u662f\uff0c\u8fd9\u7bc7\u8bba\u6587\u7684\u6807\u9898\u662f <strong>SWE-MAML<\/strong>\uff0c\u8fd9\u6697\u793a\u4e86\u539f\u59cb\u7684 MAML \u53ef\u80fd\u8fd8\u6709\u4e0d\u8db3\u4e4b\u5904\u3002<\/p>\n\n\n\n<p>\u6839\u636e\u8bba\u6587\u7b2c3\u9875\uff08\"Model-Agnostic Meta-Learning (MAML) is...\" \u9644\u8fd1\uff09\uff0c\u4f5c\u8005\u6307\u51fa\u4e86\u539f\u59cb MAML \u7684\u54ea\u4e9b<strong>\u7f3a\u70b9<\/strong>\u6216<strong>\u4e0d\u8db3<\/strong>\uff1f<\/p>\n\n\n\n<blockquote class=\"wp-block-quote is-layout-flow wp-block-quote-is-layout-flow\">\n<p>However, it comes with the need for costly hyperparameter tuning for training stability, and its performance has fallen behind many recent algorithms nowadays.<\/p>\n\n\n\n<p>Trans:\u7136\u800c\uff0c\u8fd9\u79cd\u65b9\u6cd5\u9700\u8981\u8fdb\u884c\u6602\u8d35\u7684\u8d85\u53c2\u6570\u8c03\u4f18\u4ee5\u786e\u4fdd\u8bad\u7ec3\u7a33\u5b9a\u6027\uff0c\u4e14\u5176\u6027\u80fd\u5df2\u843d\u540e\u4e8e\u5f53\u524d\u4f17\u591a\u65b0\u5174\u7b97\u6cd5\u3002<\/p>\n<\/blockquote>\n\n\n\n<p>\u8fd9\u6b63\u662f\u8bba\u6587\u6307\u51fa\u7684MAML\u7684\u4e24\u5927\u6838\u5fc3\u95ee\u9898\uff1a<\/p>\n\n\n\n<ol start=\"1\" class=\"wp-block-list\">\n<li><strong>\u8bad\u7ec3\u4e0d\u7a33\u5b9a<\/strong>\uff1a\u9700\u8981\u82b1\u8d39\u5927\u91cf\u7cbe\u529b\uff08\u6602\u8d35\u7684\u8d85\u53c2\u6570\u8c03\u4f18\uff09\u624d\u80fd\u8ba9\u5b83\u7a33\u5b9a\u8bad\u7ec3\u3002<\/li>\n\n\n\n<li><strong>\u6027\u80fd\u843d\u540e<\/strong>\uff1a\u5b83\u7684\u8bc6\u522b\u51c6\u786e\u7387\u5df2\u7ecf\u88ab\u4e00\u4e9b\u66f4\u65b0\u7684\u7b97\u6cd5\u8d85\u8d8a\u4e86\u3002<\/li>\n<\/ol>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<p>\u597d\u4e86\uff0c\u5230\u8fd9\u91cc\u6211\u4eec\u5df2\u7ecf\u5b8c\u6574\u5730\u7406\u89e3\u4e86\u201c<strong>\u4e3a\u4ec0\u4e48<\/strong>\u201d\uff08Why\uff09\u8fd9\u90e8\u5206\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>\u6838\u5fc3\u95ee\u9898<\/strong>\uff1a\u5728\u73b0\u5b9e\u519c\u4e1a\u4e2d\uff0c\u75c5\u5bb3\u6570\u636e\u5f88\u96be\u5927\u89c4\u6a21\u6536\u96c6\uff08\u73af\u5883\u590d\u6742\u3001\u4e13\u5bb6\u6807\u6ce8\u6602\u8d35\uff09\u3002<\/li>\n\n\n\n<li><strong>\u89e3\u51b3\u65b9\u5411<\/strong>\uff1a\u56e0\u6b64\u9700\u8981\u201c\u5c0f\u6837\u672c\u5b66\u4e60\u201d\uff08FSL\uff09\u6280\u672f\uff0c\u5373\u7528\u5c11\u91cf\u6570\u636e\u8bad\u7ec3\u6a21\u578b\u3002<\/li>\n\n\n\n<li><strong>\u5df2\u6709\u5de5\u5177<\/strong>\uff1aMAML \u662f\u4e00\u79cd\u6d41\u884c\u7684FSL\u65b9\u6cd5\uff0c\u5b83\u901a\u8fc7\u5b66\u4e60\u201c\u5feb\u901f\u9002\u5e94\u201d\u7684\u521d\u59cb\u5316\u53c2\u6570\u6765\u5de5\u4f5c\u3002<\/li>\n\n\n\n<li><strong>\u5de5\u5177\u7f3a\u9677<\/strong>\uff1a\u4f46 MAML \u81ea\u8eab\u5b58\u5728\u8bad\u7ec3\u6602\u8d35\u4e14\u6027\u80fd\u843d\u540e\u7684\u95ee\u9898\u3002<\/li>\n<\/ul>\n\n\n\n<p>\u8fd9\u5c31\u5f15\u51fa\u4e86\u4e00\u4e2a\u975e\u5e38\u81ea\u7136\u7684\u95ee\u9898\uff1a<strong>\u4f5c\u8005\u6253\u7b97\u5982\u4f55\u6539\u8fdb MAML\uff1f<\/strong><\/p>\n\n\n\n<p>\u8fd9\u7bc7\u8bba\u6587\u7684\u6807\u9898\uff08SWE-MAML\uff09\u548c\u6458\u8981\uff08Abstract\uff09\u5176\u5b9e\u5df2\u7ecf\u7ed9\u51fa\u4e86\u7ebf\u7d22\u3002\u4f5c\u8005\u5c06 MAML \u4e0e\u53e6\u4e00\u79cd\u5f3a\u5927\u7684\u673a\u5668\u5b66\u4e60\u8303\u5f0f\u7ed3\u5408\u4e86\u8d77\u6765\u3002<\/p>\n\n\n\n<p>\u6839\u636e\u8bba\u6587\u6458\u8981\uff08\u7b2c1\u9875\uff09\uff0c\u4f5c\u8005\u5c06 MAML \u4e0e\u4ec0\u4e48\u6280\u672f\u7ed3\u5408\u8d77\u6765\uff0c\u63d0\u51fa\u4e86\u4ed6\u4eec\u7684\u65b0\u6846\u67b6\uff1f<sup><\/sup><\/p>\n\n\n\n<blockquote class=\"wp-block-quote is-layout-flow wp-block-quote-is-layout-flow\">\n<p>This method integrates ensemble learning with Model\u0002-Agnostic Meta-Learning (MAML), allowing the effective training of multiple classifiers within the MAML framework.<\/p>\n\n\n\n<p>\u8be5\u65b9\u6cd5\u5c06\u96c6\u6210\u5b66\u4e60\u4e0e\u6a21\u578b\u65e0\u5173\u7684\u5143\u5b66\u4e60\uff08MAML\uff09\u76f8\u7ed3\u5408\uff0c\u53ef\u5728MAML\u6846\u67b6\u5185\u6709\u6548\u8bad\u7ec3\u591a\u4e2a\u5206\u7c7b\u5668\u3002<\/p>\n<\/blockquote>\n\n\n\n<p>\u6458\u8981\u91cc\u5199\u5f97\u5f88\u6e05\u695a\uff0c\u4ed6\u4eec\u63d0\u51fa\u4e86\u4e00\u79cd\u65b0\u7684\u65b9\u6cd5 <strong>SWE-MAML<\/strong>\uff0c\u5b83\u5c06 <strong>\u96c6\u6210\u5b66\u4e60\uff08ensemble learning\uff09<\/strong> \u548c MAML \u7ed3\u5408\u4e86\u8d77\u6765 <sup><\/sup>\u3002<\/p>\n\n\n\n<p>\u8fd9\u4e2a\u6846\u67b6\u7684\u8bbe\u8ba1\u76ee\u6807\uff0c\u5c31\u662f\u5728 MAML \u6846\u67b6\u5185\u6709\u6548\u8bad\u7ec3\u591a\u4e2a\u5206\u7c7b\u5668\uff08\u8bba\u6587\u79f0\u4e4b\u4e3a\u201c\u57fa\u5b66\u4e60\u5668\u201d\uff0cbase learners\uff09\u3002<\/p>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity is-style-wide\"\/>\n\n\n\n<h2 class=\"wp-block-heading has-text-align-center\">2.\u5982\u4f55 HOW<\/h2>\n\n\n\n<p>\u6211\u4eec\u5df2\u7ecf\u77e5\u9053\uff0cSWE-MAML \u7684\u6838\u5fc3\u662f\u201c<strong>\u96c6\u6210\u5b66\u4e60 + MAML<\/strong>\u201d\u3002<\/p>\n\n\n\n<p>\u6765\u770b\u770b\u8bba\u6587\u7684\u7b2c6\u9875\uff0c<strong>2.2.2 \u7ae0\u8282\uff08Sequence-weighted ensemble MAML\uff09<\/strong>\uff0c\u7279\u522b\u662f\u7b2c7\u9875\u7684 <strong>\u56fe4 (Figure 4)<\/strong>\u3002<\/p>\n\n\n<div class=\"wp-block-image\">\n<figure class=\"aligncenter size-full\"><img loading=\"lazy\" decoding=\"async\" width=\"679\" height=\"351\" src=\"https:\/\/www.leexinghai.com\/aic\/wp-content\/uploads\/2025\/11\/image-1.png\" alt=\"\" class=\"wp-image-3828\" srcset=\"https:\/\/www.leexinghai.com\/aic\/wp-content\/uploads\/2025\/11\/image-1.png 679w, https:\/\/www.leexinghai.com\/aic\/wp-content\/uploads\/2025\/11\/image-1-300x155.png 300w\" sizes=\"auto, (max-width: 679px) 100vw, 679px\" \/><\/figure>\n<\/div>\n\n\n<ul class=\"wp-block-list\">\n<li>\u6807\u51c6\u7684 MAML \u53ea\u6709\u4e00\u4e2a\u6a21\u578b\uff08\u6216\u5b66\u4e60\u5668\uff09\u3002\u4f46\u5728\u8fd9\u7bc7\u8bba\u6587\u7684\u65b9\u6cd5\u4e2d\uff0c\u56fe4 \u5411\u6211\u4eec\u5c55\u793a\u4e86\u4ec0\u4e48\u7ed3\u6784\uff1f\u8fd9\u4e2a\u96c6\u6210\uff08ensemble\uff09\u662f\u7531\u4ec0\u4e48\u7ec4\u6210\u7684\uff1f\n<ul class=\"wp-block-list\">\n<li>\u56fe4\u5c55\u793a\u4e86\u7ec4\u5408\u591a\u4e86\u4e2a\u6a21\u578b\u6765\u63d0\u5347\u5b66\u4e60\u6548\u679c\u3002\u601d\u60f3\u662f\u4e3a\u540c\u4e00\u4e2a\u4efb\u52a1\u8bad\u7ec3\u591a\u4e2a\u5b66\u4e60\u5668\uff0c\u7136\u540e\u5c06\u8fd9\u4e9b\u5b66\u4e60\u5668\u7684\u8f93\u51fa\u7ed3\u679c\u5408\u5e76\u4e3a\u6700\u7ec8\u7ed3\u679c<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n\n\n\n<p>\u8fd9\u4e2a\u65b9\u6cd5\u7684\u540d\u5b57\u91cc\u6709\u201c<strong>\u5e8f\u5217<\/strong>\u201d\uff08Sequence\uff09\u8fd9\u4e2a\u8bcd\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u4ed4\u7ec6\u770b\u56fe4\uff08\u7b2c7\u9875\uff09\u4e2d BL<sub>1<\/sub>, BL<sub>2<\/sub>, ... BL<sub>N_T<\/sub>\u4e4b\u95f4\u7684\u7bad\u5934\uff0c\u4f60\u8ba4\u4e3a\u8fd9\u4e9b\u5b66\u4e60\u5668\u662f<strong>\u5982\u4f55<\/strong>\u88ab\u8bad\u7ec3\u7684\uff1f\u5b83\u4eec\u662f\u540c\u65f6\u72ec\u7acb\u8bad\u7ec3\u7684\uff0c\u8fd8\u662f\u6709\u5148\u540e\u987a\u5e8f\uff1f\n<ul class=\"wp-block-list\">\n<li>\u6709\u5148\u540e\u987a\u5e8f<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n\n\n\n<p>\u4ece\u56fe4\u4e2d\u7684\u7bad\u5934\u53ef\u4ee5\u6e05\u695a\u5730\u770b\u5230\uff0c\u5b83\u4eec\u662f<strong>\u6309\u987a\u5e8f<\/strong>\u8bad\u7ec3\u7684\u3002\u8fd9\u5c31\u662f\u65b9\u6cd5\u540d\u4e2d\u201c<strong>\u5e8f\u5217<\/strong>\u201d\uff08Sequence\uff09\u7684\u6765\u6e90\u3002<\/p>\n\n\n\n<p>\u73b0\u5728\u6211\u4eec\u6765\u770b\u201c<strong>\u52a0\u6743<\/strong>\u201d\uff08Weighted\uff09\u8fd9\u4e2a\u8bcd\u3002<\/p>\n\n\n\n<p>\u518d\u6b21\u89c2\u5bdf\u56fe4\uff0c\u4f60\u4f1a\u53d1\u73b0\uff0c\u9664\u4e86\u6709 BL<sub>1<\/sub>, BL<sub>2<\/sub> \u8fd9\u6837\u7684\u201c\u57fa\u5b66\u4e60\u5668\u201d\u4e4b\u5916\uff0c\u8fd8\u6709\u4e00\u7ec4\u4e1c\u897f\u88ab\u7528\u6765\u8ba1\u7b97\u6700\u7ec8\u7684\u201cScore\u201d\uff08\u5f97\u5206\uff09\uff0c\u5e76\u4e14\u5b83\u4eec\u4e5f\u4f1a\u88ab\u201cupdate\u201d\uff08\u66f4\u65b0\uff09\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u4f60\u80fd\u627e\u5230\u8fd9\u7ec4\u4e1c\u897f\u662f\u4ec0\u4e48\u5417\uff1f\n<ul class=\"wp-block-list\">\n<li>\u662f w<sub>1<\/sub>, w<sub>2<\/sub>, ... w<sub>N_T<\/sub><\/li>\n<\/ul>\n<\/li>\n<\/ul>\n\n\n\n<p>\u8fd9\u4e9b\u5c31\u662f\u201c<strong>\u52a0\u6743<\/strong>\u201d\uff08Weighted\uff09\u7684\u6765\u6e90\u3002\u5982\u56fe4\u6240\u793a\uff0c\u6a21\u578b\u6700\u7ec8\u7684\u201cScore\u201d\uff08\u5206\u7c7b\u5f97\u5206\uff09\u662f\u6240\u6709\u57fa\u5b66\u4e60\u5668\uff08BLi\u200b\uff09\u7684\u8f93\u51fa\uff0c\u518d\u7528\u8fd9\u4e9b w<sub>i<\/sub>\u200b \u8fdb\u884c\u52a0\u6743\u6c42\u548c\u5f97\u5230\u7684 \u3002<\/p>\n\n\n\n<p>\u6240\u4ee5\uff0c<strong>SWE-MAML<\/strong> \u7684\u6838\u5fc3\u673a\u5236\u5c31\u662f\uff1a<\/p>\n\n\n\n<ol start=\"1\" class=\"wp-block-list\">\n<li><strong>\u5e8f\u5217\uff08Sequence\uff09<\/strong>\uff1a\u6309\u987a\u5e8f\u8bad\u7ec3\u4e00\u7cfb\u5217\u57fa\u5b66\u4e60\u5668\uff08BL<sub>1<\/sub>\u200b,BL<sub>2<\/sub>\u200b...\uff09\u3002<\/li>\n\n\n\n<li><strong>\u52a0\u6743\uff08Weighted\uff09<\/strong>\uff1a\u540c\u65f6\u5b66\u4e60\u4e00\u4e2a\u5bf9\u5e94\u7684\u6743\u91cd\uff08w<sub>1<\/sub>\u200b,w<sub>2<\/sub>\u200b...\uff09\u3002<\/li>\n\n\n\n<li><strong>\u96c6\u6210\uff08Ensemble\uff09<\/strong>\uff1a\u6700\u7ec8\u7684\u9884\u6d4b\u7ed3\u679c\u662f\u6240\u6709\u5b66\u4e60\u5668 BLi\u200b \u7684\u52a0\u6743 wi\u200b \u603b\u548c\u3002<\/li>\n<\/ol>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<p>\u73b0\u5728\uff0c\u6211\u4eec\u628a\u6700\u540e\u4e00\u5757\u62fc\u56fe\u201c<strong>MAML<\/strong>\u201d\u653e\u56de\u6765\u3002<\/p>\n\n\n\n<p>\u6211\u4eec\u5df2\u7ecf\u77e5\u9053\uff0c\u8fd9\u662f\u4e00\u4e2a\u201c\u5e8f\u5217\u201d\u8fc7\u7a0b\u3002\u90a3\u4e48\uff0c\u8fd9\u4e2a\u5e8f\u5217\u7684\u201c<strong>\u8d77\u70b9<\/strong>\u201d\uff0c\u4e5f\u5c31\u662f\u7b2c\u4e00\u4e2a\u57fa\u5b66\u4e60\u5668 BL<sub>1<\/sub>\u200b\uff0c\u5b83\u7684\u53c2\u6570\u662f<strong>\u5982\u4f55\u521d\u59cb\u5316<\/strong>\u7684\u5462\uff1f<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u6839\u636e\u56fe4\u548c\u7b2c7\u9875\u7684\u7b97\u6cd5\u63cf\u8ff0\uff08Algorithm 1\uff09\uff0c\u4f60\u80fd\u627e\u5230 BL<sub>1<\/sub>\u200b \u662f\u4ece\u54ea\u91cc\u83b7\u5f97\u5b83\u7684\u521d\u59cb\u53c2\u6570 \u03b8<sub>1<\/sub>\u200b \u7684\u5417\uff1f\n<ul class=\"wp-block-list\">\n<li>\u662f\u9996\u5148\u5b9a\u4e49\u4e86\u4e00\u4e2a\u53c2\u6570\u5316\u51fd\u6570f<sub>\u03b8<\/sub>\u8868\u793a\u7684\u6a21\u578b\uff0c\u8fd9\u4e2a\u51fd\u6570\u5305\u542b\u53c2\u6570\u03b8<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n\n\n\n<blockquote class=\"wp-block-quote is-layout-flow wp-block-quote-is-layout-flow\">\n<p>Figure 3 shows how MAML performs meta-training. In the figure, \u03b8 represents the meta-learner of the model, and f<sub>\u03b8<\/sub> is its parameterization function.<\/p>\n\n\n\n<p>Trans:\u56fe3\u5c55\u793a\u4e86MAML\u5982\u4f55\u8fdb\u884c\u5143\u8bad\u7ec3\u3002\u5176\u4e2d\uff0cq\u8868\u793a\u6a21\u578b\u7684\u5143\u5b66\u4e60\u5668\uff0cfq\u662f\u5176\u53c2\u6570\u5316\u51fd\u6570\u3002<\/p>\n\n\n\n<p>In Figure 4, \u03b8 represents the meta-learner ML. We create a set of base-learners with the same network structure of ML, denoted as BL<sub>i<\/sub>, and the decision weight of each base-learner as w<sub>i<\/sub>.<\/p>\n\n\n\n<p>Trans:\u56fe4\u4e2d\uff0c\u03b8\u4ee3\u8868\u5143\u5b66\u4e60\u5668ML\u3002\u6211\u4eec\u6784\u5efa\u4e86\u4e00\u7ec4\u5177\u6709\u76f8\u540c\u7f51\u7edc\u7ed3\u6784\u7684\u57fa\u5b66\u4e60\u5668\uff08\u8bb0\u4e3aBL<sub>i<\/sub>\uff09\uff0c\u6bcf\u4e2a\u57fa\u5b66\u4e60\u5668\u7684\u51b3\u7b56\u6743\u91cd\u4e3aw<sub>i<\/sub>\u3002<\/p>\n<\/blockquote>\n\n\n\n<p><\/p>\n\n\n\n<p>\u03b8 \u4ee3\u8868\u7684\u5c31\u662f\u201c\u5143\u5b66\u4e60\u5668\u201d\uff08Meta-Learner, ML\uff09\u7684\u53c2\u6570 \u3002<\/p>\n\n\n\n<p>\u73b0\u5728\uff0c\u8bf7\u518d\u770b\u4e00\u4e0b<strong>\u56fe4<\/strong>\u3002\u4f1a\u770b\u5230\u4e00\u4e2a\u7bad\u5934\u4ece \u03b8 \u51fa\u53d1\uff0c\u5f84\u76f4\u6307\u5411\u4e86 BL<sub>1<\/sub>\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u90a3\u4e48\uff0c\u6839\u636e\u8fd9\u4e2a\u56fe\u793a\u548c\u7b2c7\u9875\u7684\u7b97\u6cd5\u63cf\u8ff0\uff08Algorithm 1\uff09\uff0c\u6211\u4eec\u662f\u7528\u4e86\u4ec0\u4e48\u4f5c\u4e3a\u7b2c\u4e00\u4e2a\u57fa\u5b66\u4e60\u5668 BL<sub>1<\/sub> \u7684<strong>\u8d77\u59cb\u53c2\u6570<\/strong>\u5462\uff1f\n<ul class=\"wp-block-list\">\n<li>\u5c31\u662f \u03b8\uff01<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n\n\n\n<p>\u5728<strong>\u56fe4<\/strong>\u4e2d\uff0c\u6709\u4e00\u4e2a\u7bad\u5934\u4ece \u03b8\uff08\u5143\u5b66\u4e60\u5668 ML\uff09\u6307\u5411 BL<sub>1<\/sub>\uff08\u7b2c\u4e00\u4e2a\u57fa\u5b66\u4e60\u5668\uff09\u3002<\/p>\n\n\n\n<p>\u8fd9\u8bf4\u660e<strong>BL<sub>1<\/sub> \u7684\u521d\u59cb\u53c2\u6570\uff0c\u5c31\u662f\u7528\u5143\u5b66\u4e60\u5668 \u03b8 \u6765\u8bbe\u7f6e\u7684<\/strong>\u3002<\/p>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<p>\u6211\u4eec\u73b0\u5728\u77e5\u9053\u4e86\u201c\u5e8f\u5217\u201d\u7684\u8d77\u70b9 BL<sub>1<\/sub> \u662f\u5982\u4f55\u521d\u59cb\u5316\u7684\u3002<\/p>\n\n\n\n<p>\u90a3\u4e48\uff0c\u8fd9\u4e2a\u201c\u5e8f\u5217\u201d\u662f\u5982\u4f55\u5f80\u4e0b\u4f20\u9012\u7684\u5462\uff1f\u6211\u4eec\u6765\u770b\u770b BL<sub>2<\/sub> \u662f\u5982\u4f55\u521d\u59cb\u5316\u7684\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u6839\u636e<strong>\u56fe4<\/strong>\u4e2d BL<sub>1<\/sub> \u548c BL<sub>2<\/sub> \u4e4b\u95f4\u7684\u7bad\u5934\uff0c\u4ee5\u53ca\u7b2c7\u9875\u7684\u7b97\u6cd5\u63cf\u8ff0\uff08Algorithm 1\uff09\uff0c\u4f60\u8ba4\u4e3a BL<sub>2<\/sub> \u7684\u521d\u59cb\u53c2\u6570\u662f\u6765\u81ea\u54ea\u91cc\uff1f\uff08\u662f\u6765\u81ea BL<sub>1<\/sub> \u8fd8\u662f \u03b8 \uff1f\uff09\n<ul class=\"wp-block-list\">\n<li>\u6765\u81ea BL<sub>1<\/sub><\/li>\n<\/ul>\n<\/li>\n<\/ul>\n\n\n\n<p>\u4f60\u770b\u56fe4\u5c31\u660e\u767d\u4e86\uff1a<\/p>\n\n\n\n<ol start=\"1\" class=\"wp-block-list\">\n<li><strong>\u5143\u5b66\u4e60\u5668 \u03b8<\/strong> \u521d\u59cb\u5316\u4e86 <strong>BL<sub>1<\/sub><\/strong>\u3002<\/li>\n\n\n\n<li><strong><strong>BL<sub>1<\/sub><\/strong><\/strong>\uff08\u8bad\u7ec3\u540e\u7684\u53c2\u6570\uff09\u53c8\u521d\u59cb\u5316\u4e86 <strong><strong>BL<sub>1<\/sub><\/strong><\/strong>\u3002<\/li>\n\n\n\n<li>\u8fd9\u4e2a\u8fc7\u7a0b\u4f1a\u4e00\u76f4\u6301\u7eed\u4e0b\u53bb\uff0cBL<sub>i<\/sub>\u521d\u59cb\u5316 BL<sub>i+1<\/sub>\u3002<\/li>\n<\/ol>\n\n\n\n<p>\u8fd9\u5c31\u662f\u201c<strong>\u5e8f\u5217<\/strong>\u201d\uff08Sequence\uff09\u7684\u771f\u6b63\u542b\u4e49\uff1a\u5b83\u4eec\u662f\u4e00\u4e2a\u63a5\u4e00\u4e2a\u201c\u63a5\u529b\u201d\u8bad\u7ec3\u7684\u3002<\/p>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<p>\u6211\u4eec\u73b0\u5728\u628a SWE-MAML \u7684\u5de5\u4f5c\u6d41\u7a0b\uff08\u201c\u5982\u4f55\u201d\uff09\u7406\u6e05\u695a\u4e86\uff1a<\/p>\n\n\n\n<ol start=\"1\" class=\"wp-block-list\">\n<li><strong>\u8d77\u70b9<\/strong>\uff1a\u5143\u5b66\u4e60\u5668 \u03b8\u521d\u59cb\u5316 <strong><strong>BL<sub>1<\/sub><\/strong><\/strong>\u3002<\/li>\n\n\n\n<li><strong>\u5e8f\u5217<\/strong>\uff1a<strong><strong>BL<sub>1<\/sub><\/strong><\/strong> \u8bad\u7ec3\u540e\u521d\u59cb\u5316 <strong><strong>BL<sub>2<\/sub><\/strong><\/strong>\uff0c<strong><strong>BL<sub>2<\/sub><\/strong><\/strong> \u8bad\u7ec3\u540e\u521d\u59cb\u5316 <strong><strong>BL<sub>3<\/sub><\/strong><\/strong>... \u4f9d\u6b64\u7c7b\u63a8\uff0c\u8bad\u7ec3\u51fa\u4e00\u7cfb\u5217 <strong><strong>BL<sub>i<\/sub><\/strong><\/strong>\u3002<\/li>\n\n\n\n<li><strong>\u52a0\u6743<\/strong>\uff1a\u6a21\u578b\u540c\u65f6\u5b66\u4e60\u4e00\u5957\u6743\u91cd w<sub>i<\/sub>\u3002<\/li>\n\n\n\n<li><strong>\u96c6\u6210<\/strong>\uff1a\u6700\u7ec8\u7684\u9884\u6d4b\u7ed3\u679c\u662f\u6240\u6709 BL<sub>i<\/sub> \u7684\u52a0\u6743\u603b\u548c\u3002<\/li>\n\n\n\n<li><strong>MAML\uff08\u5143\u5b66\u4e60\uff09<\/strong>\uff1a\u6700\u540e\uff0c\u6a21\u578b\u4f1a\u8ba1\u7b97\u4e00\u4e2a\u201c\u5143\u635f\u5931\u201d\uff08Meta-Loss\uff09\uff0c\u7528\u8fd9\u4e2a\u635f\u5931\u53bb<strong>\u540c\u65f6\u66f4\u65b0<\/strong> \u03b8 \u548c w\u3002<\/li>\n<\/ol>\n\n\n\n<p>\u8fd9\u4e2a\u8bbe\u8ba1\u975e\u5e38\u5de7\u5999\uff0c\u5b83\u4e0d\u662f\u50cf\u4f20\u7edf\u96c6\u6210\u90a3\u6837\u8bad\u7ec3\u5b8c\u6a21\u578bA\u3001\u518d\u8bad\u7ec3\u6a21\u578bB...\u6700\u540e\u518d\u7ec4\u5408\u3002<\/p>\n\n\n\n<p>\u800c\u662f\u5728<strong>\u5143\u5b66\u4e60\u7684\u6bcf\u4e00\u6b65<\/strong>\u90fd\u540c\u65f6\u8bad\u7ec3\u8fd9\u4e2a\u201c\u5e8f\u5217\u96c6\u6210\u201d\u548c\u201c\u6743\u91cd\u201d\uff0c\u8ba9\u6574\u4e2a\u7cfb\u7edf\u5b66\u4f1a<strong>\u5982\u4f55\u5feb\u901f\u6784\u5efa\u4e00\u4e2a\u5f3a\u5927\u7684\u96c6\u6210\u56e2\u961f<\/strong>\u3002<\/p>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<p>\u6211\u4eec\u5df2\u7ecf\u5f04\u6e05\u4e86\u201c\u4e3a\u4ec0\u4e48\u201d\uff08Why\uff09\u548c\u201c\u5982\u4f55\u201d\uff08How\uff09\u3002<\/p>\n\n\n\n<p>\u63a5\u4e0b\u6765\uff0c\u6211\u4eec\u81ea\u7136\u8981\u5173\u5fc3\uff1a<strong>\u201c\u6548\u679c\u5982\u4f55\uff1f\u201d\uff08How Well?\uff09<\/strong><\/p>\n\n\n\n<p>\u8fd9\u4e2a\u65b0\u65b9\u6cd5 SWE-MAML \u542c\u8d77\u6765\u66f4\u590d\u6742\u4e86\uff0c\u5b83\u771f\u7684\u6bd4\u539f\u59cb\u7684 MAML \u548c\u5176\u4ed6\u65b9\u6cd5\u66f4\u597d\u5417\uff1f<\/p>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity is-style-wide\"\/>\n\n\n\n<h2 class=\"wp-block-heading has-text-align-center\">3.\u6548\u679c\u5982\u4f55HOW WELL<\/h2>\n\n\n\n<p>\u8fd9\u7bc7\u8bba\u6587\u7684\u4e00\u4e2a\u6838\u5fc3\u521b\u65b0\u662f\u201c\u96c6\u6210\u201d\uff08ensemble\uff09\uff0c\u4e5f\u5c31\u662f\u4f7f\u7528\u591a\u4e2a\u57fa\u5b66\u4e60\u5668\uff08base-learners\uff09\u3002\u4e00\u4e2a\u5f88\u81ea\u7136\u7684\u95ee\u9898\u662f\uff1a<strong>\u57fa\u5b66\u4e60\u5668\u7684\u6570\u91cf<\/strong>\u4f1a\u5982\u4f55\u5f71\u54cd\u6a21\u578b\u7684\u6027\u80fd\uff1f<\/p>\n\n\n\n<p>\u4f5c\u8005\u5728 3.2 \u8282 \u548c <strong>\u56fe6 (\u7b2c10\u9875)<\/strong> \u4e2d\u63a2\u8ba8\u4e86\u8fd9\u4e2a\u95ee\u9898\u3002<\/p>\n\n\n\n<p>\u8bba\u6587\u4e2d\u63d0\u5230\uff0c\u4ed6\u4eec\u5c06\u539f\u59cb\u7684 MAML \u89c6\u4e3a\u4e00\u4e2a\u7279\u4f8b\u3002\u4f60\u80fd\u627e\u5230 MAML \u88ab\u5f53\u4f5c\u6709\u51e0\u4e2a\u57fa\u5b66\u4e60\u5668\u7684\u60c5\u51b5\u5417\uff1f<sup><\/sup><\/p>\n\n\n\n<figure class=\"wp-block-image size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"428\" src=\"https:\/\/www.leexinghai.com\/aic\/wp-content\/uploads\/2025\/11\/image-2-1024x428.png\" alt=\"\" class=\"wp-image-3831\" srcset=\"https:\/\/www.leexinghai.com\/aic\/wp-content\/uploads\/2025\/11\/image-2-1024x428.png 1024w, https:\/\/www.leexinghai.com\/aic\/wp-content\/uploads\/2025\/11\/image-2-300x125.png 300w, https:\/\/www.leexinghai.com\/aic\/wp-content\/uploads\/2025\/11\/image-2-768x321.png 768w, https:\/\/www.leexinghai.com\/aic\/wp-content\/uploads\/2025\/11\/image-2.png 1332w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<blockquote class=\"wp-block-quote is-layout-flow wp-block-quote-is-layout-flow\">\n<p>For comparison, we treat MAML as a special case of SWE-MAML where the number of base-learner is <strong>only 1<\/strong>. The results are shown in Figure 6, in which BL_num indicates the number of base-learners and the error bars represents the 95% confidence intervals.<\/p>\n\n\n\n<p>Trans:\u4e3a\u4e86\u8fdb\u884c\u6bd4\u8f83\uff0c\u6211\u4eec\u5c06 MAML \u89c6\u4e3a SWE-MAML \u7684\u4e00\u4e2a\u7279\u4f8b\uff0c\u5176\u4e2d\u57fa\u672c\u5b66\u4e60\u5668\u7684\u6570\u91cf<strong>\u4ec5\u4e3a 1<\/strong>\u3002\u7ed3\u679c\u5982\u56fe 6 \u6240\u793a\uff0c\u5176\u4e2d BL_num \u8868\u793a\u57fa\u672c\u5b66\u4e60\u5668\u7684\u6570\u91cf\uff0c\u8bef\u5dee\u7ebf\u8868\u793a 95% \u7f6e\u4fe1\u533a\u95f4\u3002<\/p>\n<\/blockquote>\n\n\n\n<p>\u4f5c\u8005\u662f\u5c06<strong>\u539f\u59cb\u7684 MAML<\/strong> \u89c6\u4e3a\u4e00\u4e2a\u57fa\u5b66\u4e60\u5668\u6570\u91cf\u4e3a<strong>1<\/strong>\u7684\u7279\u4f8b <sup><\/sup>\u3002<\/p>\n\n\n\n<p>\u73b0\u5728\uff0c\u8ba9\u6211\u4eec\u4e00\u8d77\u770b\u770b<strong>\u56fe6A<\/strong>\uff08\u7b2c10\u9875\uff09\u3002\u8fd9\u5f20\u56fe\u6807\u51fa\u4e86\u57fa\u5b66\u4e60\u5668\uff08BL_num\uff09\u6570\u91cf\u4ece1\u52307\u65f6\u7684\u51c6\u786e\u7387\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u5f53 BL<sub>num<\/sub> = 1 (\u4e5f\u5c31\u662f\u539f\u59cb MAML) \u65f6\uff0c\u5728 \"5-way 1-shot\" (\u7ea2\u8272\u5b9e\u7ebf) \u4efb\u52a1\u4e0a\uff0c\u51c6\u786e\u7387\u5927\u6982\u662f\u591a\u5c11\uff1f<\/li>\n\n\n\n<li>\u5f53 BL<sub>num<\/sub> \u589e\u52a0\u5230 5 \u6216 6 \u65f6\uff0c\u51c6\u786e\u7387\u53c8\u53d8\u6210\u4e86\u591a\u5c11\uff1f<\/li>\n<\/ul>\n\n\n\n<p>\u5728 \"5-way 1-shot\" (\u7ea2\u8272\u5b9e\u7ebf) \u4efb\u52a1\u4e0a\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u5f53 BL<sub>num<\/sub> = 1 (\u539f\u59cb MAML) \u65f6\uff0c\u51c6\u786e\u7387\u5927\u7ea6\u662f <strong>66%<\/strong> \u3002<\/li>\n\n\n\n<li>\u5f53 BL<sub>num<\/sub> = 5 \u65f6\uff0c\u51c6\u786e\u7387\u8fbe\u5230\u4e86 <strong>74.56%<\/strong> \u3002<\/li>\n\n\n\n<li>\u5f53 BL<sub>num<\/sub> = 6 \u65f6\uff0c\u51c6\u786e\u7387\u662f 73.29% \u3002<\/li>\n<\/ul>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<p>\u5728 \"5-way 5-shot\" (\u7eff\u8272\u865a\u7ebf) \u4efb\u52a1\u4e0a\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u5f53 BL<sub>num<\/sub> = 1 (\u539f\u59cb MAML) \u65f6\uff0c\u51c6\u786e\u7387\u5927\u7ea6\u662f <strong>85.8%<\/strong> \u3002<\/li>\n\n\n\n<li>\u5f53 BL<sub>num<\/sub> = 6 \u65f6\uff0c\u51c6\u786e\u7387\u8fbe\u5230\u4e86 <strong>89.75%<\/strong> \u3002<\/li>\n<\/ul>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<p>\u6211\u4eec\u4ece\u8fd9\u4e9b\u6570\u636e\u4e2d\u5f97\u51fa\u4e86\u4e00\u4e2a\u975e\u5e38\u6e05\u6670\u7684\u7ed3\u8bba\uff1a<strong>SWE-MAML \u663e\u8457\u4f18\u4e8e \u539f\u59cb\u7684 MAML<\/strong>\u3002<\/p>\n\n\n\n<p>\u4ec5\u5728 \"5-way 1-shot\" \u4efb\u52a1\u4e0a\uff0c\u51c6\u786e\u7387\u5c31\u63d0\u5347\u4e86\u8d85\u8fc7 <strong>8.5%<\/strong> (\u4ece 66% \u5230 74.56%) <sup><\/sup><sup><\/sup><sup><\/sup><sup><\/sup>\u3002<\/p>\n\n\n\n<p>\u8fd9\u8bc1\u660e\u4e86\u201c\u96c6\u6210\u201d\u8fd9\u4e2a\u601d\u8def\u662f\u6709\u6548\u7684\u3002<\/p>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<p>\u4f5c\u8005\u8fd8\u505a\u4e86\u7b2c\u4e8c\u4e2a\u5b9e\u9a8c\u6765\u9a8c\u8bc1\u8fd9\u4e2a\u6a21\u578b\u3002\u5728 3.4 \u8282 \u548c <strong>\u88685 (Table 5)<\/strong> \u4e2d\uff0c\u4ed6\u4eec\u5c06 SWE-MAML \u4e0e<strong>\u5176\u4ed6\u591a\u79cd FSL \u7b97\u6cd5<\/strong>\u8fdb\u884c\u4e86\u6bd4\u8f83\u3002<\/p>\n\n\n\n<p>\u901a\u8fc7\u89c2\u5bdf\u88685\uff08\u7b2c12\u9875\uff09\uff0c\u53ef\u4ee5\u77e5\u9053 SWE-MAML\uff08\u6700\u540e\u4e00\u884c\uff09\u4e0e ProtoNet\u3001MatchingNet\u3001RelationNet \u8fd9\u4e9b\u7ecf\u5178\u65b9\u6cd5\u76f8\u6bd4\uff0c\u7ed3\u679c\u5982\u4f55\uff1f<\/p>\n\n\n\n<figure class=\"wp-block-image size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"634\" src=\"https:\/\/www.leexinghai.com\/aic\/wp-content\/uploads\/2025\/11\/image-3-1024x634.png\" alt=\"\" class=\"wp-image-3832\" srcset=\"https:\/\/www.leexinghai.com\/aic\/wp-content\/uploads\/2025\/11\/image-3-1024x634.png 1024w, https:\/\/www.leexinghai.com\/aic\/wp-content\/uploads\/2025\/11\/image-3-300x186.png 300w, https:\/\/www.leexinghai.com\/aic\/wp-content\/uploads\/2025\/11\/image-3-768x475.png 768w, https:\/\/www.leexinghai.com\/aic\/wp-content\/uploads\/2025\/11\/image-3.png 1344w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p>\u5728 <strong>5-way 1-shot<\/strong> (Conv4) \u4efb\u52a1\u4e0a\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>SWE-MAML<\/strong> \u662f <strong>74.56%<\/strong> <\/li>\n\n\n\n<li><strong>ProtoNet<\/strong> \u662f <strong>68.61%<\/strong> <\/li>\n<\/ul>\n\n\n\n<p>\u8fd9\u786e\u5b9e\u662f\u4e00\u4e2a\u63a5\u8fd16\u4e2a\u767e\u5206\u70b9\u7684\u663e\u8457\u63d0\u5347\u3002<\/p>\n\n\n\n<p>\u5728 <strong>5-way 5-shot<\/strong> (Conv4) \u4efb\u52a1\u4e0a\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>SWE-MAML<\/strong> \u662f <strong>89.75%<\/strong><\/li>\n\n\n\n<li><strong>ProtoNet<\/strong> \u662f <strong>83.84%<\/strong><\/li>\n<\/ul>\n\n\n\n<p>\u8fd9\u53c8\u662f\u4e00\u4e2a\u63a5\u8fd16\u4e2a\u767e\u5206\u70b9\u7684\u63d0\u5347\u3002<\/p>\n\n\n\n<p>\u5982\u679c\u518d\u7eb5\u89c2 <strong>\u88685<\/strong> (\u7b2c12\u9875) \u7684\u6240\u6709\u884c\uff0c\u5c31\u53ef\u4ee5\u53d1\u73b0\u4e00\u4e2a\u6e05\u6670\u7684\u6a21\u5f0f\uff1a\u65e0\u8bba\u662f\u5728 5-way 1-shot \u8fd8\u662f 5-way 5-shot \u4efb\u52a1\u4e0a\uff0c\u4e5f\u65e0\u8bba\u4f7f\u7528 Conv4 \u8fd8\u662f Conv6\uff0cSWE-MAML\uff08\u6700\u540e\u4e00\u884c\uff09\u7684\u51c6\u786e\u7387<strong>\u603b\u662f<\/strong>\u6392\u5728\u7b2c\u4e00\u4f4d \u3002<\/p>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<p>\u6211\u4eec\u76ee\u524d\u5df2\u7ecf\u9a8c\u8bc1\u4e86\uff1a<\/p>\n\n\n\n<ol start=\"1\" class=\"wp-block-list\">\n<li><strong>\u201c\u96c6\u6210\u201d\u6709\u6548<\/strong>\uff1a\u4f7f\u7528\u591a\u4e2a\u57fa\u5b66\u4e60\u5668 (BL) \u663e\u8457\u4f18\u4e8e MAML (BL=1) \u3002<\/li>\n\n\n\n<li><strong>\u201c\u6027\u80fd\u9886\u5148\u201d<\/strong>\uff1aSWE-MAML \u5728\u6807\u51c6\u6570\u636e\u96c6 (PlantVillage) \u4e0a\u7684\u8868\u73b0\u4f18\u4e8e\u5176\u4ed6\u6240\u6709\u88ab\u6bd4\u8f83\u7684 FSL \u7b97\u6cd5 \u3002<\/li>\n<\/ol>\n\n\n\n<p>\u8fd9\u770b\u8d77\u6765\u975e\u5e38\u68d2\u3002\u4f46\u4f5c\u8005\u4eec\u8fdb\u884c\u4e86\u4e00\u9879<strong>\u6700\u7ec8\u6d4b\u8bd5<\/strong>\uff0c\u53ef\u4ee5\u8bf4\u662f\u5bf9\u6a21\u578b\u771f\u6b63\u7684\u8003\u9a8c\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u5728 3.5 \u8282 (\u7b2c12\u9875)\uff0c\u4ed6\u4eec\u8bbe\u8ba1\u4e86\u4e00\u4e2a\u201c<strong>\u573a\u666fB<\/strong>\u201d\uff08Scenario B\uff09 \u3002\u8fd8\u8bb0\u5f97\u8fd9\u4e2a\u573a\u666f\u4e0e\u6211\u4eec\u4e4b\u524d\u770b\u7684\u201c\u573a\u666fA\u201d\u6709\u4ec0\u4e48\u5173\u952e\u533a\u522b\u5417\uff1f\uff08\u63d0\u793a\uff1a\u4e0e\u6570\u636e\u96c6\u6709\u5173\uff09\n<ul class=\"wp-block-list\">\n<li>\u573a\u666fA\u662f\u539f\u57df\u548c\u76ee\u6807\u57df\u90fd\u662fPlantVillage\uff0c\u800c\u573a\u666fB\u662f\u539f\u57df\u662fPlantVillage\uff0c\u76ee\u6807\u57df\u662fPDD\u3002<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n\n\n\n<p>PDD \u6307\u7684\u662f <strong>Potato Disease Dataset<\/strong>\uff08\u9a6c\u94c3\u85af\u75c5\u5bb3\u6570\u636e\u96c6\uff09 <sup><\/sup>\u3002<\/p>\n\n\n\n<p>\u573a\u666fA\u548c\u573a\u666fB\u7684\u5173\u952e\u533a\u522b\u5728\u4e8e\u6d4b\u8bd5\u7528\u7684<strong>\u76ee\u6807\u57df<\/strong>\u4e0d\u540c\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>\u573a\u666fA<\/strong>\uff1a\u8bad\u7ec3\u548c\u6d4b\u8bd5\u90fd\u6765\u81ea PlantVillage \u6570\u636e\u96c6 \u3002<\/li>\n\n\n\n<li><strong>\u573a\u666fB<\/strong>\uff1a\u8bad\u7ec3\u6765\u81ea PlantVillage\uff0c\u4f46\u6d4b\u8bd5\u6765\u81ea PDD \u3002<\/li>\n<\/ul>\n\n\n\n<p>\u8fd9\u7bc7\u8bba\u6587\u7279\u522b\u6307\u51fa\uff0cPDD \u4e0e PlantVillage \u6709\u4e00\u4e2a<strong>\u672c\u8d28\u7684\u533a\u522b<\/strong>\uff0c\u8fd9\u4f7f\u5f97\u573a\u666fB\u7684\u6d4b\u8bd5\u53d8\u5f97\u5c24\u5176\u56f0\u96be\u548c\u6709\u610f\u4e49\u3002<\/p>\n\n\n\n<blockquote class=\"wp-block-quote is-layout-flow wp-block-quote-is-layout-flow\">\n<p>The second dataset is the Potato Disease Dataset (PDD). Unlike the PlantVillage dataset, all PDD images are captured under natural lighting conditions with complex backgrounds.<\/p>\n\n\n\n<p>Trans:\u7b2c\u4e8c\u4e2a\u6570\u636e\u96c6\u662f\u9a6c\u94c3\u85af\u75c5\u5bb3\u6570\u636e\u96c6\uff08PDD\uff09\u3002\u4e0ePlantVillage\u6570\u636e\u96c6\u4e0d\u540c\uff0c\u6240\u6709PDD\u56fe\u50cf\u5747\u5728\u81ea\u7136\u5149\u7167\u6761\u4ef6\u4e0b\u62cd\u6444\uff0c\u80cc\u666f\u590d\u6742\u3002<\/p>\n<\/blockquote>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u6839\u636e\u8bba\u6587 2.1 \u8282\uff08\u7b2c5\u9875\uff09\u5bf9\u8fd9\u4e24\u4e2a\u6570\u636e\u96c6\u7684\u63cf\u8ff0\uff0c\u4f60\u8ba4\u4e3a PDD \u56fe\u50cf\u7684\u4e3b\u8981\u7279\u70b9\u662f\u4ec0\u4e48\uff1f\u5b83\u548c PlantVillage \u56fe\u50cf\uff08\u7b2c4\u9875\uff0c\u56fe1\uff09\u76f8\u6bd4\uff0c\u6700\u5927\u7684\u4e0d\u540c\u5728\u54ea\u91cc\uff1f\n<ul class=\"wp-block-list\">\n<li><strong>PlantVillage<\/strong>\uff1a\u662f\u5728\u5b9e\u9a8c\u5ba4\u6761\u4ef6\u4e0b\u62cd\u6444\u7684\uff0c\u80cc\u666f\u7b80\u5355\uff0c\u5149\u7167\u53d7\u63a7 \u3002<\/li>\n\n\n\n<li><strong>PDD<\/strong>\uff1a\u662f\u5728<strong>\u81ea\u7136\u6761\u4ef6<\/strong>\u4e0b\u62cd\u6444\u7684\uff0c\u80cc\u666f\u975e\u5e38\u590d\u6742\uff08\u6bd4\u5982\u6709\u5176\u4ed6\u53f6\u5b50\u3001\u571f\u58e4\u3001\u5149\u5f71\uff09\uff0c\u5149\u7167\u4e5f\u4e0d\u5747\u5300 \u3002<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n\n\n\n<p>\u8fd9\u5c31\u662fAI\u9886\u57df\u6240\u8bf4\u7684\u201c<strong>\u57df\u5dee\u5f02<\/strong>\u201d\uff08Domain Shift\uff09\u3002\u6a21\u578b\u5728\u201c\u5e72\u51c0\u201d\u7684\u5b9e\u9a8c\u5ba4\u6570\u636e\uff08\u6e90\u57df\uff09\u4e0a\u8bad\u7ec3\uff0c\u5374\u8981\u5728\u201c\u6df7\u4e71\u201d\u7684\u771f\u5b9e\u4e16\u754c\u6570\u636e\uff08\u76ee\u6807\u57df\uff09\u4e0a\u6d4b\u8bd5\u3002<\/p>\n\n\n\n<p>\u8fd9\u5bf9\u6a21\u578b\u6765\u8bf4\u662f\u4e00\u4e2a\u5de8\u5927\u7684\u8003\u9a8c\u3002\u5728\u573a\u666fA\u4e2d\uff0c\u6a21\u578b\u53ea\u662f\u8bc6\u522b\u5b83\u6ca1\u89c1\u8fc7\u7684<em>\u75c5\u5bb3<\/em>\uff0c\u4f46\u56fe\u50cf\u7684<em>\u98ce\u683c<\/em>\u662f\u76f8\u4f3c\u7684\u3002\u800c\u5728\u573a\u666fB\u4e2d\uff0c\u8fde\u56fe\u50cf\u7684<strong>\u98ce\u683c\u90fd\u5b8c\u5168\u4e0d\u540c\u4e86<\/strong>\u3002<\/p>\n\n\n\n<p>\u8fd9\u5c31\u50cf\u4f60\u53ea\u5b66\u8fc7\u5370\u5237\u4f53\u7684\u5b57\u6bcd\uff0c\u73b0\u5728\u5374\u8981\u4f60\u53bb\u8bc6\u522b\u6f66\u8349\u7684\u624b\u5199\u4f53\u3002<\/p>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<p>\u90a3\u4e48\uff0c\u9762\u5bf9\u8fd9\u4e2a\u6781\u5177\u6311\u6218\u6027\u7684\u201c\u573a\u666fB\u201d\uff0cSWE-MAML \u548c\u5176\u4ed6\u7b97\u6cd5\u7684\u8868\u73b0\u5982\u4f55\u5462\uff1f<\/p>\n\n\n\n<p>\u8ba9\u6211\u4eec\u6765\u770b\u770b\u8bba\u6587\u7684<strong>\u88686 (Table 6)<\/strong>\uff08\u7b2c13\u9875\uff09\u3002\u8fd9\u5f20\u8868\u663e\u793a\u4e86\u6a21\u578b\u5728PDD\u6570\u636e\u96c6\u4e0a\u7684\u51c6\u786e\u7387\u3002<\/p>\n\n\n\n<p>\u8bf7\u770b SWE-MAML (\u6700\u540e\u4e00\u884c) \u548c\u5176\u4ed6\u6240\u6709\u65b9\u6cd5\uff08\u4e0a\u9762\u51e0\u884c\uff09\u7684\u5bf9\u6bd4\u7ed3\u679c\u3002\u4f60\u5f97\u51fa\u4e86\u4ec0\u4e48\u7ed3\u8bba\uff1f<\/p>\n\n\n\n<figure class=\"wp-block-image size-large\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"460\" src=\"https:\/\/www.leexinghai.com\/aic\/wp-content\/uploads\/2025\/11\/image-4-1024x460.png\" alt=\"\" class=\"wp-image-3834\" srcset=\"https:\/\/www.leexinghai.com\/aic\/wp-content\/uploads\/2025\/11\/image-4-1024x460.png 1024w, https:\/\/www.leexinghai.com\/aic\/wp-content\/uploads\/2025\/11\/image-4-300x135.png 300w, https:\/\/www.leexinghai.com\/aic\/wp-content\/uploads\/2025\/11\/image-4-768x345.png 768w, https:\/\/www.leexinghai.com\/aic\/wp-content\/uploads\/2025\/11\/image-4.png 1340w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p>\u5728<strong>\u88686<\/strong>\uff08\u7b2c13\u9875\uff09\u4e2d\uff0c\u65e0\u8bba\u662f\u5728\u54ea\u4e2a \"shot\"\uff081-shot, 5-shot, 10-shot...\uff09\u7684\u8bbe\u7f6e\u4e0b\uff0cSWE-MAML\uff08\u6700\u540e\u4e00\u884c\uff09\u7684\u51c6\u786e\u7387\u90fd<strong>\u59cb\u7ec8\u662f\u6700\u9ad8<\/strong>\u7684 <sup><\/sup>\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u5728\u201c1-shot\u201d\u65f6\uff0cSWE-MAML \u8fbe\u5230\u4e86 <strong>39.82%<\/strong>\uff0c\u800c\u5176\u4ed6\u65b9\u6cd5\u90fd\u5728 39% \u4ee5\u4e0b\u3002<\/li>\n\n\n\n<li>\u5728\u201c30-shot\u201d\u65f6\uff0cSWE-MAML \u8fbe\u5230\u4e86 <strong>75.71%<\/strong>\uff0c\u800c\u7b2c\u4e8c\u540d\uff08DeepEMD-FCN\uff09\u53ea\u6709 74.70% \u3002<\/li>\n<\/ul>\n\n\n\n<p>\u8fd9\u4e2a\u7ed3\u679c\u610f\u4e49\u91cd\u5927\u3002\u5b83\u8868\u660e\uff0c\u5373\u4f7f\u9762\u5bf9\u201c\u57df\u5dee\u5f02\u201d\uff08\u5b9e\u9a8c\u5ba4 vs. \u81ea\u7136\u73af\u5883\uff09\u8fd9\u4e2a\u4e25\u5cfb\u7684\u6311\u6218\uff0cSWE-MAML \u4f9d\u7136\u6bd4\u6240\u6709\u5176\u4ed6\u65b9\u6cd5\u66f4\u7a33\u5065\u3001\u66f4\u51c6\u786e\u3002<\/p>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<p>\u6211\u4eec\u73b0\u5728\u5df2\u7ecf\u5b8c\u6210\u4e86\u5bf9\u201c<strong>\u6548\u679c\u5982\u4f55<\/strong>\u201d\u7684\u5206\u6790\uff1a<\/p>\n\n\n\n<ol start=\"1\" class=\"wp-block-list\">\n<li><strong>\u4f18\u4e8e MAML<\/strong>\uff1a\u96c6\u6210\u591a\u4e2a\u57fa\u5b66\u4e60\u5668 (5-7\u4e2a) \u663e\u8457\u4f18\u4e8e MAML (1\u4e2a) \u3002<\/li>\n\n\n\n<li><strong>\u4f18\u4e8e\u5176\u4ed6 FSL<\/strong>\uff1a\u5728\u6807\u51c6\u6570\u636e\u96c6 (PlantVillage) \u4e0a\uff0cSWE-MAML \u8868\u73b0\u6700\u4f73 \u3002<\/li>\n\n\n\n<li><strong>\u8de8\u57df\u4f9d\u7136\u6709\u6548<\/strong>\uff1a\u5728\u6781\u5177\u6311\u6218\u7684\u201c\u5b9e\u9a8c\u5ba4\u5230\u81ea\u7136\u73af\u5883\u201d\u6d4b\u8bd5 (PDD) \u4e2d\uff0cSWE-MAML \u4f9d\u7136\u4fdd\u6301\u7b2c\u4e00 \u3002<\/li>\n<\/ol>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<p>\u6211\u4eec\u5df2\u7ecf\u4e00\u8d77\u5b66\u4e60\u4e86\u8fd9\u7bc7\u8bba\u6587\u7684\u201c\u4e3a\u4ec0\u4e48\u201d\uff08Why\uff09\u3001\u201c\u5982\u4f55\u201d\uff08How\uff09\u4ee5\u53ca\u201c\u6548\u679c\u5982\u4f55\u201d\uff08How Well\uff09\u3002<\/p>\n\n\n\n<p>\u6700\u540e\uff0c\u6211\u4eec\u770b\u770b<strong>\u7b2c4\u8282 \u8ba8\u8bba\uff08Discussion\uff09<\/strong> \u548c<strong>\u7b2c5\u8282 \u7ed3\u8bba\uff08Conclusion\uff09<\/strong>\uff0c\u6765\u603b\u7ed3\u4e00\u4e0b\u4f5c\u8005\u8ba4\u4e3a\u4ed6\u4eec\u8fd9\u9879\u5de5\u4f5c\u7684<strong>\u6838\u5fc3\u8d21\u732e<\/strong>\u548c<strong>\u5c40\u9650\u6027<\/strong>\u3002<\/p>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity is-style-dots\"\/>\n\n\n\n<h2 class=\"wp-block-heading has-text-align-center\">4.\u8ba8\u8bba DISCUSSION<\/h2>\n\n\n\n<p>\u4f5c\u8005\u5728\u8fd9\u4e00\u8282\u56de\u987e\u4e86\u4ed6\u4eec\u7684\u4e3b\u8981\u5b9e\u9a8c\u6210\u679c\uff0c\u5e76\u5f3a\u8c03\u4e86\u5b83\u4eec\u7684\u610f\u4e49\u3002<\/p>\n\n\n\n<p>\u4f5c\u8005\u7279\u522b\u63d0\u5230\u4e86\u4e00\u4e2a\u57283.3\u8282\uff08\u56fe7\uff09\u4e2d\u4e5f\u63a2\u8ba8\u8fc7\u7684\u91cd\u8981\u56e0\u7d20\u3002\u9664\u4e86\u201c\u96c6\u6210\u5b66\u4e60\u5668\u7684\u6570\u91cf\u201d\uff08\u6211\u4eec\u4e4b\u524d\u5728\u56fe6\u770b\u5230\u7684\uff09\u4e4b\u5916\uff0c\u4f5c\u8005\u8fd8\u53d1\u73b0\u4e86<strong>\u4ec0\u4e48\u56e0\u7d20<\/strong>\u5bf9\u6a21\u578b\u7684\u8bc6\u522b\u51c6\u786e\u7387\u6709\u201c\u66f4\u663e\u8457\u7684\u5f71\u54cd\u201d\uff08a more significant effect\uff09\uff1f <sup><\/sup><\/p>\n\n\n\n<blockquote class=\"wp-block-quote is-layout-flow wp-block-quote-is-layout-flow\">\n<p>The number of disease classes involved in the source domain has a more significant effect on recognition accuracy. <\/p>\n\n\n\n<p>Trans:\u6e90\u57df\u4e2d\u6d89\u53ca\u7684\u75be\u75c5\u7c7b\u522b\u6570\u91cf\u5bf9\u8bc6\u522b\u51c6\u786e\u5ea6\u7684\u5f71\u54cd\u66f4\u4e3a\u663e\u8457\u3002<\/p>\n<\/blockquote>\n\n\n\n<p>\u8bba\u6587\u5728\u7b2c4\u8282\u201c\u8ba8\u8bba\u201d <sup><\/sup> \u548c\u7b2c5\u8282\u201c\u7ed3\u8bba\u201d\u4e2d\u90fd\u5f3a\u8c03\u4e86\u8fd9\u4e00\u70b9\uff1a<\/p>\n\n\n\n<blockquote class=\"wp-block-quote is-layout-flow wp-block-quote-is-layout-flow\">\n<p>This reveals the fact that in the meta-learning phase, diverse disease categories help improve the accuracy of the model. <\/p>\n\n\n\n<p>Trans:\u5728\u5143\u5b66\u4e60\u9636\u6bb5\uff08meta-learning phase\uff09\uff0c<strong>\u591a\u6837\u5316\u7684\u75be\u75c5\u7c7b\u522b<\/strong>\u6709\u52a9\u4e8e\u63d0\u9ad8\u6a21\u578b\u7684\u51c6\u786e\u6027\u3002<\/p>\n<\/blockquote>\n\n\n\n<p>\u4f5c\u8005\u6307\u51fa\uff0c\u5f53\u6e90\u57df\uff08source domain\uff09\u4e2d\u7684\u9884\u8bad\u7ec3\u7c7b\u522b\u4ece11\u4e2a\u589e\u52a0\u523027\u4e2a\u65f6\uff1a<\/p>\n\n\n\n<blockquote class=\"wp-block-quote is-layout-flow wp-block-quote-is-layout-flow\">\n<p>In both 5-way 1-shot and 5-way 5-shot conditions, the recognition accuracy of the model improved by 12.01% and 8.45%.<\/p>\n<\/blockquote>\n\n\n\n<ul class=\"wp-block-list\">\n<li>5-way 1-shot \u7684\u51c6\u786e\u7387\u63d0\u5347\u4e86 <strong>12.01%<\/strong> <\/li>\n\n\n\n<li>5-way 5-shot \u7684\u51c6\u786e\u7387\u63d0\u5347\u4e86 <strong>8.45%<\/strong> <\/li>\n<\/ul>\n\n\n\n<p>\u8fd9\u8bc1\u660e\u4e86\u5728\u5143\u8bad\u7ec3\u9636\u6bb5\u7ed9\u6a21\u578b\u770b\uff08\u201c\u5b66\u4e60\u201d\uff09\u7684<strong>\u79cd\u7c7b<\/strong>\u8d8a\u591a\uff0c\u6a21\u578b\u5b66\u4e60\u201c\u5982\u4f55\u5b66\u4e60\u201d\u7684\u80fd\u529b\u5c31\u8d8a\u5f3a\uff0c\u5f53\u5b83\u9047\u5230\u4e00\u4e2a\u5168\u65b0\u7684\u75c5\u5bb3\u65f6\uff0c\u5b83\u4e5f\u80fd\u8bc6\u522b\u5f97\u66f4\u597d\u3002<\/p>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<p>\u6211\u4eec\u5df2\u7ecf\u5b8c\u6574\u5730\u5b66\u4e60\u4e86\u8fd9\u7bc7\u8bba\u6587\u3002\u6211\u4eec\u77e5\u9053\u4e86\uff1a<\/p>\n\n\n\n<ol start=\"1\" class=\"wp-block-list\">\n<li><strong>\u4e3a\u4ec0\u4e48\uff08Why\uff09<\/strong>\uff1a\u89e3\u51b3\u771f\u5b9e\u519c\u4e1a\u4e2d\u75c5\u5bb3\u6570\u636e\u7a00\u7f3a\u7684\u95ee\u9898 \u3002<\/li>\n\n\n\n<li><strong>\u5982\u4f55\u505a\uff08How\uff09<\/strong>\uff1a\u63d0\u51fa\u4e86 SWE-MAML\uff0c\u5c06\u201c\u5e8f\u5217\u96c6\u6210\u5b66\u4e60\u201d\u5d4c\u5165\u5230 MAML \u6846\u67b6\u4e2d \u3002<\/li>\n\n\n\n<li><strong>\u6548\u679c\u5982\u4f55\uff08How Well\uff09<\/strong>\uff1a\u5728\u6807\u51c6\u6570\u636e\u96c6 \u548c\u6781\u5177\u6311\u6218\u7684\u8de8\u57df\u6570\u636e\u96c6 (PDD)  \u4e0a\u5747\u8d85\u8d8a\u4e86\u5176\u4ed6\u65b9\u6cd5\u3002<\/li>\n\n\n\n<li><strong>\u5173\u952e\u53d1\u73b0<\/strong>\uff1a\u96c6\u6210\u5b66\u4e60\u5668\u7684\u6570\u91cf\uff085-7\u4e2a\u6700\u4f73\uff09 \u548c\u6e90\u57df\u7684\u7c7b\u522b\u6570\u91cf  \u662f\u63d0\u5347\u6027\u80fd\u7684\u5173\u952e\u3002<\/li>\n<\/ol>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity is-style-dots\"\/>\n\n\n\n<h2 class=\"wp-block-heading has-text-align-center\">5.\u7ed3\u8bbaCONCLUSION<\/h2>\n\n\n\n<p>\u8bba\u6587\u5728\u7b2c5\u8282\u201c\u7ed3\u8bba\u201d\u7684\u672b\u5c3e\u660e\u786e\u6307\u51fa\u4e86\u8fd9\u4e00\u70b9\uff1a<\/p>\n\n\n\n<blockquote class=\"wp-block-quote is-layout-flow wp-block-quote-is-layout-flow\">\n<p>Meanwhile, the model\u2019s performance needs to be improved in the cross-domain few-shot disease recognition task.<\/p>\n\n\n\n<p>Trans:\u201c\u540c\u65f6\uff0c\u8be5\u6a21\u578b\u5728<strong>\u8de8\u9886\u57df\u5c0f\u6837\u672c\u75be\u75c5\u8bc6\u522b\u4efb\u52a1<\/strong>\uff08cross-domain few-shot disease recognition task\uff09\u4e2d\u7684\u6027\u80fd\u4ecd\u9700\u63d0\u9ad8\u3002\u201d<\/p>\n<\/blockquote>\n\n\n\n<p>\u8fd9\u6b63\u662f\u4ed6\u4eec\u8ba1\u5212\u5728\u672a\u6765\u7814\u7a76\u4e2d\u91cd\u70b9\u653b\u514b\u7684\u65b9\u5411 <sup><\/sup>\u3002<\/p>\n\n\n\n<p>\u8fd9\u4e5f\u5f88\u5408\u7406\uff0c\u6211\u4eec\u5728\u5b9e\u9a8c\u7ed3\u679c\uff08\u88686\uff09\u4e2d\u4e5f\u770b\u5230\u4e86\uff0c\u5c3d\u7ba1 SWE-MAML \u5728 PDD\uff08\u81ea\u7136\u573a\u666f\uff09\u4e0a\u7684\u8868\u73b0\u5df2\u7ecf\u662f\u6700\u597d\u7684\uff0c\u4f46\u51c6\u786e\u7387<strong>\uff08\u5982 75.71%\uff09<\/strong>\u76f8\u6bd4\u4e8e\u5728 PlantVillage \u5185\u90e8\u6d4b\u8bd5<strong>\uff08\u88685\uff0c\u5982 89.75%\uff09<\/strong>\u8fd8\u662f\u6709\u660e\u663e\u4e0b\u964d\u7684\u3002\u8fd9\u8bf4\u660e<strong>\u4ece\u201c\u5b9e\u9a8c\u5ba4\u201d\u5230\u201c\u771f\u5b9e\u4e16\u754c\u201d\u7684\u201c\u8de8\u57df\u201d\u95ee\u9898<\/strong>\uff0c\u4f9d\u7136\u662f\u8fd9\u4e2a\u9886\u57df\u6700\u5927\u7684\u6311\u6218\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u672c\u671f\u8bc4\u8ff0\u6587\u7ae0\uff1a \u5b83\u63a2\u8ba8\u7684\u662f\u4e00\u4e2a\u5728\u519c\u4e1a\u548cAI\u4ea4\u53c9\u9886\u57df\u975e\u5e38\u91cd\u8981\u7684\u95ee\u9898\uff1a\u5982\u4f55\u5728\u6570\u636e\u6837\u672c\u5f88\u5c11\u7684\u60c5\u51b5\u4e0b\uff08\u5373\u201c\u5c0f\u6837\u672c\u5b66\u4e60\u201d [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":3825,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[15],"tags":[85,83],"class_list":["post-3824","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-microtalk","tag-fsl","tag-zuhui4"],"_links":{"self":[{"href":"https:\/\/www.leexinghai.com\/aic\/wp-json\/wp\/v2\/posts\/3824","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.leexinghai.com\/aic\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.leexinghai.com\/aic\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.leexinghai.com\/aic\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.leexinghai.com\/aic\/wp-json\/wp\/v2\/comments?post=3824"}],"version-history":[{"count":9,"href":"https:\/\/www.leexinghai.com\/aic\/wp-json\/wp\/v2\/posts\/3824\/revisions"}],"predecessor-version":[{"id":4040,"href":"https:\/\/www.leexinghai.com\/aic\/wp-json\/wp\/v2\/posts\/3824\/revisions\/4040"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.leexinghai.com\/aic\/wp-json\/wp\/v2\/media\/3825"}],"wp:attachment":[{"href":"https:\/\/www.leexinghai.com\/aic\/wp-json\/wp\/v2\/media?parent=3824"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.leexinghai.com\/aic\/wp-json\/wp\/v2\/categories?post=3824"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.leexinghai.com\/aic\/wp-json\/wp\/v2\/tags?post=3824"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}