{"id":477882,"date":"2023-08-09T09:22:01","date_gmt":"2023-08-09T09:22:01","guid":{"rendered":""},"modified":"2023-09-05T11:15:36","modified_gmt":"2023-09-05T11:15:36","slug":"long-short-term-memory-lstm","status":"publish","type":"wiki","link":"https:\/\/oneproxy.pro\/cn\/wiki\/long-short-term-memory-lstm\/","title":{"rendered":"\u957f\u77ed\u671f\u8bb0\u5fc6\uff08LSTM\uff09"},"content":{"rendered":"<p>\u957f\u77ed\u671f\u8bb0\u5fc6 (LSTM) \u662f\u4e00\u79cd\u4eba\u5de5\u5faa\u73af\u795e\u7ecf\u7f51\u7edc (RNN) \u67b6\u6784\uff0c\u65e8\u5728\u514b\u670d\u4f20\u7edf RNN \u5728\u6355\u6349\u5e8f\u5217\u6570\u636e\u4e2d\u7684\u957f\u671f\u4f9d\u8d56\u5173\u7cfb\u65b9\u9762\u7684\u5c40\u9650\u6027\u3002LSTM \u7684\u5f15\u5165\u662f\u4e3a\u4e86\u89e3\u51b3\u5728\u5904\u7406\u957f\u5e8f\u5217\u65f6\u963b\u788d RNN \u8bad\u7ec3\u7684\u6d88\u5931\u548c\u7206\u70b8\u68af\u5ea6\u95ee\u9898\u3002\u5b83\u5e7f\u6cdb\u5e94\u7528\u4e8e\u5404\u4e2a\u9886\u57df\uff0c\u5305\u62ec\u81ea\u7136\u8bed\u8a00\u5904\u7406\u3001\u8bed\u97f3\u8bc6\u522b\u3001\u65f6\u95f4\u5e8f\u5217\u9884\u6d4b\u7b49\u3002<\/p>\n<h2>\u957f\u77ed\u671f\u8bb0\u5fc6\uff08LSTM\uff09\u7684\u8d77\u6e90\u5386\u53f2\u4ee5\u53ca\u9996\u6b21\u63d0\u53ca\u5b83<\/h2>\n<p>LSTM \u67b6\u6784\u6700\u521d\u7531 Sepp Hochreiter \u548c J\u00fcrgen Schmidhuber \u4e8e 1997 \u5e74\u63d0\u51fa\u3002\u4ed6\u4eec\u7684\u8bba\u6587\u300a\u957f\u77ed\u671f\u8bb0\u5fc6\u300b\u5f15\u5165\u4e86 LSTM \u5355\u5143\u7684\u6982\u5ff5\uff0c\u4ee5\u89e3\u51b3\u4f20\u7edf RNN \u9762\u4e34\u7684\u95ee\u9898\u3002\u4ed6\u4eec\u8bc1\u660e\u4e86 LSTM \u5355\u5143\u53ef\u4ee5\u6709\u6548\u5730\u5b66\u4e60\u548c\u4fdd\u7559\u5e8f\u5217\u4e2d\u7684\u957f\u671f\u4f9d\u8d56\u5173\u7cfb\uff0c\u4f7f\u5176\u975e\u5e38\u9002\u5408\u6d89\u53ca\u590d\u6742\u65f6\u95f4\u6a21\u5f0f\u7684\u4efb\u52a1\u3002<\/p>\n<h2>\u5173\u4e8e\u957f\u77ed\u671f\u8bb0\u5fc6 (LSTM) \u7684\u8be6\u7ec6\u4fe1\u606f<\/h2>\n<p>LSTM \u662f\u57fa\u672c RNN \u6a21\u578b\u7684\u6269\u5c55\uff0c\u5177\u6709\u66f4\u590d\u6742\u7684\u5185\u90e8\u7ed3\u6784\uff0c\u4f7f\u5176\u80fd\u591f\u9009\u62e9\u6027\u5730\u957f\u671f\u4fdd\u7559\u6216\u5fd8\u8bb0\u4fe1\u606f\u3002LSTM \u80cc\u540e\u7684\u6838\u5fc3\u601d\u60f3\u662f\u4f7f\u7528\u8bb0\u5fc6\u5355\u5143\uff0c\u8bb0\u5fc6\u5355\u5143\u662f\u8d1f\u8d23\u968f\u65f6\u95f4\u5b58\u50a8\u548c\u66f4\u65b0\u4fe1\u606f\u7684\u5355\u5143\u3002\u8fd9\u4e9b\u8bb0\u5fc6\u5355\u5143\u7531\u4e09\u4e2a\u4e3b\u8981\u7ec4\u4ef6\u63a7\u5236\uff1a\u8f93\u5165\u95e8\u3001\u5fd8\u8bb0\u95e8\u548c\u8f93\u51fa\u95e8\u3002<\/p>\n<h3>\u957f\u77ed\u671f\u8bb0\u5fc6 (LSTM) \u7684\u5de5\u4f5c\u539f\u7406<\/h3>\n<ol>\n<li>\n<p><strong>\u8f93\u5165\u95e8\uff1a<\/strong> \u8f93\u5165\u95e8\u63a7\u5236\u6709\u591a\u5c11\u65b0\u4fe1\u606f\u88ab\u6dfb\u52a0\u5230\u8bb0\u5fc6\u5355\u5143\u3002\u5b83\u4ece\u5f53\u524d\u65f6\u95f4\u6b65\u9aa4\u83b7\u53d6\u8f93\u5165\uff0c\u5e76\u51b3\u5b9a\u54ea\u4e9b\u90e8\u5206\u4e0e\u5b58\u50a8\u5728\u8bb0\u5fc6\u4e2d\u6709\u5173\u3002<\/p>\n<\/li>\n<li>\n<p><strong>\u5fd8\u8bb0\u95e8\uff1a<\/strong> \u9057\u5fd8\u95e8\u51b3\u5b9a\u54ea\u4e9b\u4fe1\u606f\u9700\u8981\u4ece\u8bb0\u5fc6\u5355\u5143\u4e2d\u4e22\u5f03\u3002\u5b83\u4ece\u524d\u4e00\u4e2a\u65f6\u95f4\u6b65\u9aa4\u548c\u5f53\u524d\u65f6\u95f4\u6b65\u9aa4\u83b7\u53d6\u8f93\u5165\uff0c\u5e76\u51b3\u5b9a\u524d\u4e00\u4e2a\u8bb0\u5fc6\u7684\u54ea\u4e9b\u90e8\u5206\u4e0d\u518d\u76f8\u5173\u3002<\/p>\n<\/li>\n<li>\n<p><strong>\u8f93\u51fa\u95e8\uff1a<\/strong> \u8f93\u51fa\u95e8\u8c03\u8282\u4ece\u8bb0\u5fc6\u5355\u5143\u4e2d\u63d0\u53d6\u5e76\u7528\u4f5c LSTM \u5355\u5143\u8f93\u51fa\u7684\u4fe1\u606f\u91cf\u3002<\/p>\n<\/li>\n<\/ol>\n<p>\u901a\u8fc7\u8fd9\u4e9b\u95e8\u6765\u8c03\u8282\u4fe1\u606f\u6d41\u7684\u80fd\u529b\u4f7f\u5f97 LSTM \u80fd\u591f\u7ef4\u6301\u957f\u671f\u4f9d\u8d56\u6027\uff0c\u5e76\u514b\u670d\u4f20\u7edf RNN \u9762\u4e34\u7684\u68af\u5ea6\u6d88\u5931\u548c\u7206\u70b8\u95ee\u9898\u3002<\/p>\n<h2>\u957f\u77ed\u671f\u8bb0\u5fc6\uff08LSTM\uff09\u7684\u5173\u952e\u7279\u5f81\u5206\u6790<\/h2>\n<p>LSTM \u5177\u6709\u51e0\u4e2a\u5173\u952e\u7279\u6027\uff0c\u4f7f\u5176\u6210\u4e3a\u5904\u7406\u5e8f\u5217\u6570\u636e\u7684\u6709\u6548\u5de5\u5177\uff1a<\/p>\n<ul>\n<li>\n<p><strong>\u957f\u671f\u4f9d\u8d56\u6027\uff1a<\/strong> LSTM \u53ef\u4ee5\u6355\u83b7\u5e76\u8bb0\u4f4f\u9065\u8fdc\u7684\u8fc7\u53bb\u65f6\u95f4\u6b65\u9aa4\u7684\u4fe1\u606f\uff0c\u4f7f\u5176\u975e\u5e38\u9002\u5408\u5177\u6709\u8fdc\u8ddd\u79bb\u4f9d\u8d56\u6027\u7684\u4efb\u52a1\u3002<\/p>\n<\/li>\n<li>\n<p><strong>\u907f\u514d\u68af\u5ea6\u95ee\u9898\uff1a<\/strong> LSTM \u7684\u67b6\u6784\u6709\u52a9\u4e8e\u7f13\u89e3\u68af\u5ea6\u6d88\u5931\u548c\u7206\u70b8\u95ee\u9898\uff0c\u4ece\u800c\u786e\u4fdd\u66f4\u7a33\u5b9a\u3001\u66f4\u9ad8\u6548\u7684\u8bad\u7ec3\u3002<\/p>\n<\/li>\n<li>\n<p><strong>\u9009\u62e9\u6027\u8bb0\u5fc6\uff1a<\/strong> LSTM \u5355\u5143\u53ef\u4ee5\u6709\u9009\u62e9\u5730\u5b58\u50a8\u548c\u5fd8\u8bb0\u4fe1\u606f\uff0c\u4ece\u800c\u4f7f\u5b83\u4eec\u80fd\u591f\u4e13\u6ce8\u4e8e\u8f93\u5165\u5e8f\u5217\u4e2d\u6700\u76f8\u5173\u7684\u65b9\u9762\u3002<\/p>\n<\/li>\n<li>\n<p><strong>\u591a\u529f\u80fd\u6027\uff1a<\/strong> LSTM \u53ef\u4ee5\u5904\u7406\u4e0d\u540c\u957f\u5ea6\u7684\u5e8f\u5217\uff0c\u4f7f\u5176\u80fd\u591f\u9002\u5e94\u5404\u79cd\u5b9e\u9645\u5e94\u7528\u3002<\/p>\n<\/li>\n<\/ul>\n<h2>\u957f\u77ed\u671f\u8bb0\u5fc6\uff08LSTM\uff09\u7684\u7c7b\u578b<\/h2>\n<p>LSTM \u968f\u7740\u65f6\u95f4\u7684\u63a8\u79fb\u4e0d\u65ad\u53d1\u5c55\uff0c\u5f62\u6210\u4e86\u4e0d\u540c\u7684\u53d8\u4f53\u548c\u6269\u5c55\u3002\u4ee5\u4e0b\u662f\u4e00\u4e9b\u503c\u5f97\u6ce8\u610f\u7684 LSTM \u7c7b\u578b\uff1a<\/p>\n<ol>\n<li>\n<p><strong>\u539f\u59cb LSTM\uff1a<\/strong> \u524d\u9762\u63cf\u8ff0\u7684\u6807\u51c6 LSTM \u67b6\u6784\u3002<\/p>\n<\/li>\n<li>\n<p><strong>\u95e8\u63a7\u5faa\u73af\u5355\u5143\uff08GRU\uff09\uff1a<\/strong> \u4ec5\u5177\u6709\u4e24\u4e2a\u95e8\uff08\u91cd\u7f6e\u95e8\u548c\u66f4\u65b0\u95e8\uff09\u7684 LSTM \u7b80\u5316\u7248\u672c\u3002<\/p>\n<\/li>\n<li>\n<p><strong>\u7aa5\u5b54LSTM\uff1a<\/strong> LSTM \u7684\u6269\u5c55\uff0c\u5141\u8bb8\u95e8\u76f4\u63a5\u8bbf\u95ee\u5355\u5143\u72b6\u6001\u3002<\/p>\n<\/li>\n<li>\n<p><strong>\u5e26\u6709\u6ce8\u610f\u529b\u673a\u5236\u7684 LSTM\uff1a<\/strong> \u5c06 LSTM \u4e0e\u6ce8\u610f\u529b\u673a\u5236\u76f8\u7ed3\u5408\uff0c\u5173\u6ce8\u8f93\u5165\u5e8f\u5217\u7684\u7279\u5b9a\u90e8\u5206\u3002<\/p>\n<\/li>\n<li>\n<p><strong>\u53cc\u5411 LSTM\uff1a<\/strong> LSTM \u53d8\u4f53\uff0c\u53ef\u5411\u524d\u548c\u5411\u540e\u5904\u7406\u8f93\u5165\u5e8f\u5217\u3002<\/p>\n<\/li>\n<li>\n<p><strong>\u5806\u53e0 LSTM\uff1a<\/strong> \u4f7f\u7528\u591a\u5c42 LSTM \u5355\u5143\u6765\u6355\u83b7\u6570\u636e\u4e2d\u66f4\u590d\u6742\u7684\u6a21\u5f0f\u3002<\/p>\n<\/li>\n<\/ol>\n<h2>\u957f\u77ed\u671f\u8bb0\u5fc6\uff08LSTM\uff09\u7684\u4f7f\u7528\u65b9\u6cd5\u3001\u4f7f\u7528\u4e2d\u9047\u5230\u7684\u95ee\u9898\u53ca\u89e3\u51b3\u65b9\u6cd5<\/h2>\n<p>LSTM \u53ef\u5e94\u7528\u4e8e\u5404\u4e2a\u9886\u57df\uff0c\u5305\u62ec\uff1a<\/p>\n<ol>\n<li>\n<p><strong>\u81ea\u7136\u8bed\u8a00\u5904\u7406\uff1a<\/strong> LSTM \u7528\u4e8e\u6587\u672c\u751f\u6210\u3001\u60c5\u611f\u5206\u6790\u3001\u673a\u5668\u7ffb\u8bd1\u548c\u8bed\u8a00\u5efa\u6a21\u3002<\/p>\n<\/li>\n<li>\n<p><strong>\u8bed\u97f3\u8bc6\u522b\uff1a<\/strong> LSTM \u6709\u52a9\u4e8e\u8bed\u97f3\u5230\u6587\u672c\u7684\u8f6c\u6362\u548c\u8bed\u97f3\u52a9\u624b\u3002<\/p>\n<\/li>\n<li>\n<p><strong>\u65f6\u95f4\u5e8f\u5217\u9884\u6d4b\uff1a<\/strong> LSTM \u7528\u4e8e\u80a1\u7968\u5e02\u573a\u9884\u6d4b\u3001\u5929\u6c14\u9884\u62a5\u548c\u80fd\u6e90\u8d1f\u8377\u9884\u6d4b\u3002<\/p>\n<\/li>\n<li>\n<p><strong>\u624b\u52bf\u8bc6\u522b\uff1a<\/strong> LSTM \u53ef\u4ee5\u8bc6\u522b\u57fa\u4e8e\u624b\u52bf\u7684\u4ea4\u4e92\u4e2d\u7684\u6a21\u5f0f\u3002<\/p>\n<\/li>\n<\/ol>\n<p>\u7136\u800c\uff0cLSTM \u4e5f\u5b58\u5728\u6311\u6218\uff0c\u4f8b\u5982\uff1a<\/p>\n<ul>\n<li>\n<p><strong>\u8ba1\u7b97\u590d\u6742\u6027\uff1a<\/strong> \u8bad\u7ec3 LSTM \u6a21\u578b\u53ef\u80fd\u9700\u8981\u5927\u91cf\u8ba1\u7b97\uff0c\u5c24\u5176\u662f\u5728\u5904\u7406\u5927\u578b\u6570\u636e\u96c6\u65f6\u3002<\/p>\n<\/li>\n<li>\n<p><strong>\u8fc7\u62df\u5408\uff1a<\/strong> LSTM \u6a21\u578b\u5bb9\u6613\u8fc7\u5ea6\u62df\u5408\uff0c\u4f46\u53ef\u4ee5\u901a\u8fc7\u6b63\u5219\u5316\u6280\u672f\u548c\u66f4\u591a\u6570\u636e\u6765\u7f13\u89e3\u3002<\/p>\n<\/li>\n<li>\n<p><strong>\u8bad\u7ec3\u65f6\u95f4\u957f\uff1a<\/strong> LSTM \u8bad\u7ec3\u53ef\u80fd\u9700\u8981\u5927\u91cf\u7684\u65f6\u95f4\u548c\u8d44\u6e90\uff0c\u7279\u522b\u662f\u5bf9\u4e8e\u6df1\u5ea6\u548c\u590d\u6742\u7684\u67b6\u6784\u3002<\/p>\n<\/li>\n<\/ul>\n<p>\u4e3a\u4e86\u514b\u670d\u8fd9\u4e9b\u6311\u6218\uff0c\u7814\u7a76\u4eba\u5458\u548c\u4ece\u4e1a\u8005\u4e00\u76f4\u81f4\u529b\u4e8e\u6539\u8fdb\u4f18\u5316\u7b97\u6cd5\uff0c\u5f00\u53d1\u66f4\u9ad8\u6548\u7684\u67b6\u6784\uff0c\u5e76\u63a2\u7d22\u8fc1\u79fb\u5b66\u4e60\u6280\u672f\u3002<\/p>\n<h2>\u4e3b\u8981\u7279\u5f81\u4ee5\u53ca\u4e0e\u7c7b\u4f3c\u672f\u8bed\u7684\u5176\u4ed6\u6bd4\u8f83\u4ee5\u8868\u683c\u548c\u5217\u8868\u7684\u5f62\u5f0f<\/h2>\n<p>\u4ee5\u4e0b\u662f LSTM \u4e0e\u5176\u4ed6\u76f8\u5173\u672f\u8bed\u7684\u6bd4\u8f83\uff1a<\/p>\n<table>\n<thead>\n<tr>\n<th>\u5b66\u671f<\/th>\n<th>\u63cf\u8ff0<\/th>\n<th>\u4e3b\u8981\u5dee\u5f02<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>RNN\uff08\u5faa\u73af\u795e\u7ecf\u7f51\u7edc\uff09<\/td>\n<td>\u4e00\u79cd\u7528\u4e8e\u5904\u7406\u987a\u5e8f\u6570\u636e\u7684\u795e\u7ecf\u7f51\u7edc<\/td>\n<td>\u7f3a\u4e4f LSTM \u5904\u7406\u957f\u671f\u4f9d\u8d56\u5173\u7cfb\u7684\u80fd\u529b<\/td>\n<\/tr>\n<tr>\n<td>GRU\uff08\u95e8\u63a7\u5faa\u73af\u5355\u5143\uff09<\/td>\n<td>\u5177\u6709\u66f4\u5c11\u95e8\u9650\u7684 LSTM \u7b80\u5316\u7248\u672c<\/td>\n<td>\u66f4\u5c11\u7684\u95e8\uff0c\u66f4\u7b80\u5355\u7684\u67b6\u6784<\/td>\n<\/tr>\n<tr>\n<td>\u53d8\u538b\u5668<\/td>\n<td>\u5e8f\u5217\u5230\u5e8f\u5217\u6a21\u578b\u67b6\u6784<\/td>\n<td>\u65e0\u5faa\u73af\uff0c\u81ea\u6ce8\u610f\u529b\u673a\u5236<\/td>\n<\/tr>\n<tr>\n<td>\u5e26\u6ce8\u610f\u529b\u673a\u5236\u7684 LSTM<\/td>\n<td>\u7ed3\u5408\u6ce8\u610f\u529b\u673a\u5236\u7684LSTM<\/td>\n<td>\u589e\u5f3a\u5bf9\u8f93\u5165\u5e8f\u5217\u76f8\u5173\u90e8\u5206\u7684\u5173\u6ce8<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<h2>\u4e0e\u957f\u77ed\u671f\u8bb0\u5fc6 (LSTM) \u76f8\u5173\u7684\u672a\u6765\u89c2\u70b9\u548c\u6280\u672f<\/h2>\n<p>LSTM \u53ca\u5176\u5e94\u7528\u7684\u672a\u6765\u524d\u666f\u5149\u660e\u3002\u968f\u7740\u6280\u672f\u7684\u8fdb\u6b65\uff0c\u6211\u4eec\u53ef\u4ee5\u671f\u5f85\u4ee5\u4e0b\u9886\u57df\u5f97\u5230\u6539\u8fdb\uff1a<\/p>\n<ol>\n<li>\n<p><strong>\u6548\u7387\uff1a<\/strong> \u6b63\u5728\u8fdb\u884c\u7684\u7814\u7a76\u5c06\u96c6\u4e2d\u4e8e\u4f18\u5316 LSTM \u67b6\u6784\u4ee5\u51cf\u5c11\u8ba1\u7b97\u8981\u6c42\u548c\u8bad\u7ec3\u65f6\u95f4\u3002<\/p>\n<\/li>\n<li>\n<p><strong>\u8fc1\u79fb\u5b66\u4e60\uff1a<\/strong> \u5229\u7528\u9884\u5148\u8bad\u7ec3\u7684 LSTM \u6a21\u578b\u6765\u5b8c\u6210\u7279\u5b9a\u4efb\u52a1\uff0c\u4ee5\u63d0\u9ad8\u6548\u7387\u548c\u6cdb\u5316\u80fd\u529b\u3002<\/p>\n<\/li>\n<li>\n<p><strong>\u8de8\u5b66\u79d1\u5e94\u7528\uff1a<\/strong> LSTM \u5c06\u7ee7\u7eed\u5e94\u7528\u4e8e\u533b\u7597\u4fdd\u5065\u3001\u91d1\u878d\u548c\u81ea\u4e3b\u7cfb\u7edf\u7b49\u4e0d\u540c\u9886\u57df\u3002<\/p>\n<\/li>\n<li>\n<p><strong>\u6df7\u5408\u67b6\u6784\uff1a<\/strong> \u5c06 LSTM \u4e0e\u5176\u4ed6\u6df1\u5ea6\u5b66\u4e60\u6a21\u578b\u76f8\u7ed3\u5408\uff0c\u4ee5\u63d0\u9ad8\u6027\u80fd\u548c\u7279\u5f81\u63d0\u53d6\u3002<\/p>\n<\/li>\n<\/ol>\n<h2>\u4ee3\u7406\u670d\u52a1\u5668\u5982\u4f55\u4e0e\u957f\u77ed\u671f\u8bb0\u5fc6 (LSTM) \u4e00\u8d77\u4f7f\u7528\u6216\u5173\u8054<\/h2>\n<p>\u4ee3\u7406\u670d\u52a1\u5668\u5728\u7f51\u7edc\u6293\u53d6\u3001\u6570\u636e\u6536\u96c6\u548c\u5904\u7406\u5927\u89c4\u6a21\u6570\u636e\u6d41\u4e2d\u8d77\u7740\u81f3\u5173\u91cd\u8981\u7684\u4f5c\u7528\u3002\u4e0e LSTM \u7ed3\u5408\u4f7f\u7528\u65f6\uff0c\u4ee3\u7406\u670d\u52a1\u5668\u53ef\u4ee5\u901a\u8fc7\u591a\u79cd\u65b9\u5f0f\u5e2e\u52a9\u589e\u5f3a\u57fa\u4e8e LSTM \u7684\u6a21\u578b\u7684\u6027\u80fd\uff1a<\/p>\n<ol>\n<li>\n<p><strong>\u6570\u636e\u91c7\u96c6\uff1a<\/strong> \u4ee3\u7406\u670d\u52a1\u5668\u53ef\u4ee5\u5c06\u6570\u636e\u6536\u96c6\u4efb\u52a1\u5206\u5e03\u5728\u591a\u4e2a IP \u5730\u5740\u4e0a\uff0c\u9632\u6b62\u901f\u7387\u9650\u5236\u5e76\u786e\u4fdd LSTM \u8bad\u7ec3\u7684\u6570\u636e\u6d41\u7a33\u5b9a\u3002<\/p>\n<\/li>\n<li>\n<p><strong>\u9690\u79c1\u548c\u5b89\u5168\uff1a<\/strong> \u4ee3\u7406\u670d\u52a1\u5668\u63d0\u4f9b\u4e86\u989d\u5916\u7684\u533f\u540d\u5c42\uff0c\u4fdd\u62a4\u654f\u611f\u6570\u636e\u5e76\u786e\u4fdd\u57fa\u4e8e LSTM \u7684\u5e94\u7528\u7a0b\u5e8f\u7684\u5b89\u5168\u8fde\u63a5\u3002<\/p>\n<\/li>\n<li>\n<p><strong>\u8d1f\u8f7d\u5747\u8861\uff1a<\/strong> \u4ee3\u7406\u670d\u52a1\u5668\u6709\u52a9\u4e8e\u5728\u5904\u7406\u591a\u4e2a\u8bf7\u6c42\u65f6\u5206\u914d\u8ba1\u7b97\u8d1f\u8f7d\uff0c\u4ece\u800c\u4f18\u5316 LSTM \u6027\u80fd\u3002<\/p>\n<\/li>\n<li>\n<p><strong>\u57fa\u4e8e\u4f4d\u7f6e\u7684\u5206\u6790\uff1a<\/strong> \u4f7f\u7528\u6765\u81ea\u4e0d\u540c\u5730\u7406\u4f4d\u7f6e\u7684\u4ee3\u7406\u53ef\u4ee5\u4f7f LSTM \u6a21\u578b\u6355\u6349\u7279\u5b9a\u533a\u57df\u7684\u6a21\u5f0f\u548c\u884c\u4e3a\u3002<\/p>\n<\/li>\n<\/ol>\n<p>\u901a\u8fc7\u5c06\u4ee3\u7406\u670d\u52a1\u5668\u4e0e LSTM \u5e94\u7528\u7a0b\u5e8f\u96c6\u6210\uff0c\u7528\u6237\u53ef\u4ee5\u4f18\u5316\u6570\u636e\u91c7\u96c6\u3001\u589e\u5f3a\u5b89\u5168\u6027\u5e76\u63d0\u9ad8\u6574\u4f53\u6027\u80fd\u3002<\/p>\n<h2>\u76f8\u5173\u94fe\u63a5<\/h2>\n<p>\u6709\u5173\u957f\u77ed\u671f\u8bb0\u5fc6\uff08LSTM\uff09\u7684\u66f4\u591a\u4fe1\u606f\uff0c\u53ef\u4ee5\u53c2\u8003\u4ee5\u4e0b\u8d44\u6e90\uff1a<\/p>\n<ol>\n<li><a href=\"https:\/\/www.bioinf.jku.at\/publications\/older\/2604.pdf\" target=\"_new\" rel=\"noopener nofollow\">Hochreiter \u548c Schmidhuber \u7684\u539f\u59cb LSTM \u8bba\u6587<\/a><\/li>\n<li><a href=\"https:\/\/colah.github.io\/posts\/2015-08-Understanding-LSTMs\/\" target=\"_new\" rel=\"noopener nofollow\">\u7406\u89e3 LSTM \u7f51\u7edc \u2013 Colah \u7684\u535a\u5ba2<\/a><\/li>\n<li><a href=\"https:\/\/en.wikipedia.org\/wiki\/Long_short-term_memory\" target=\"_new\" rel=\"noopener nofollow\">\u957f\u77ed\u671f\u8bb0\u5fc6\uff08LSTM\uff09\u2014\u2014\u7ef4\u57fa\u767e\u79d1<\/a><\/li>\n<\/ol>\n<p>\u603b\u4e4b\uff0c\u957f\u77ed\u671f\u8bb0\u5fc6 (LSTM) \u5f7b\u5e95\u6539\u53d8\u4e86\u5e8f\u5217\u5efa\u6a21\u548c\u5206\u6790\u9886\u57df\u3002\u5b83\u80fd\u591f\u5904\u7406\u957f\u671f\u4f9d\u8d56\u5173\u7cfb\u5e76\u907f\u514d\u68af\u5ea6\u95ee\u9898\uff0c\u4f7f\u5176\u6210\u4e3a\u5404\u79cd\u5e94\u7528\u7684\u70ed\u95e8\u9009\u62e9\u3002\u968f\u7740\u6280\u672f\u7684\u4e0d\u65ad\u53d1\u5c55\uff0cLSTM \u6709\u671b\u5728\u5851\u9020\u4eba\u5de5\u667a\u80fd\u548c\u6570\u636e\u9a71\u52a8\u51b3\u7b56\u7684\u672a\u6765\u65b9\u9762\u53d1\u6325\u8d8a\u6765\u8d8a\u91cd\u8981\u7684\u4f5c\u7528\u3002<\/p>","protected":false},"featured_media":468808,"menu_order":0,"template":"","meta":{"_acf_changed":false,"content-type":"","inline_featured_image":false,"footnotes":""},"class_list":["post-477882","wiki","type-wiki","status-publish","has-post-thumbnail","hentry"],"acf":{"faq_title":"Frequently Asked Questions about <mark>Long Short-Term Memory (LSTM)<\/mark>","faq_items":[{"question":"What is Long Short-Term Memory (LSTM)?","answer":"<p>Long Short-Term Memory (LSTM) is a type of artificial recurrent neural network (RNN) designed to overcome the limitations of traditional RNNs in capturing long-term dependencies in sequential data. It can effectively learn and retain information from distant past time steps, making it ideal for tasks involving complex temporal patterns.<\/p>"},{"question":"Who developed LSTM and when was it first introduced?","answer":"<p>LSTM was first proposed by Sepp Hochreiter and J\u00fcrgen Schmidhuber in 1997. Their paper titled \"Long Short-Term Memory\" introduced the concept of LSTM units as a solution to the vanishing and exploding gradient problems faced by traditional RNNs.<\/p>"},{"question":"How does Long Short-Term Memory (LSTM) work?","answer":"<p>LSTM consists of memory cells with input, forget, and output gates. The input gate controls new information's addition to the memory cell, the forget gate decides what information to discard, and the output gate regulates the information extracted from the memory. This selective memory mechanism allows LSTM to capture and remember long-term dependencies.<\/p>"},{"question":"What are the key features of Long Short-Term Memory (LSTM)?","answer":"<p>The key features of LSTM include its ability to handle long-term dependencies, overcome gradient problems, selectively retain or forget information, and adapt to sequences of varying lengths.<\/p>"},{"question":"What types of Long Short-Term Memory (LSTM) exist?","answer":"<p>Various types of LSTM include Vanilla LSTM, Gated Recurrent Unit (GRU), Peephole LSTM, LSTM with Attention, Bidirectional LSTM, and Stacked LSTM. Each type has specific characteristics and applications.<\/p>"},{"question":"How can Long Short-Term Memory (LSTM) be used?","answer":"<p>LSTM finds applications in natural language processing, speech recognition, time series prediction, gesture recognition, and more. It is used for text generation, sentiment analysis, weather prediction, and stock market forecasting, among other tasks.<\/p>"},{"question":"What are the challenges related to LSTM usage, and how can they be addressed?","answer":"<p>Challenges include computational complexity, overfitting, and long training times. These issues can be mitigated through optimization algorithms, regularization techniques, and using transfer learning.<\/p>"},{"question":"How does Long Short-Term Memory (LSTM) compare to other related terms?","answer":"<p>LSTM differs from basic RNNs by its ability to capture long-term dependencies. It is more complex than Gated Recurrent Units (GRU) and lacks the self-attention mechanism of Transformers.<\/p>"},{"question":"What are the future perspectives of Long Short-Term Memory (LSTM)?","answer":"<p>The future of LSTM looks promising, with ongoing research focusing on efficiency, transfer learning, interdisciplinary applications, and hybrid architectures.<\/p>"},{"question":"How can proxy servers be associated with Long Short-Term Memory (LSTM)?","answer":"<p>Proxy servers can enhance LSTM performance by enabling efficient data collection, providing privacy and security, load balancing, and facilitating location-based analysis.<\/p>"}]},"_links":{"self":[{"href":"https:\/\/oneproxy.pro\/cn\/wp-json\/wp\/v2\/wiki\/477882","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/oneproxy.pro\/cn\/wp-json\/wp\/v2\/wiki"}],"about":[{"href":"https:\/\/oneproxy.pro\/cn\/wp-json\/wp\/v2\/types\/wiki"}],"version-history":[{"count":0,"href":"https:\/\/oneproxy.pro\/cn\/wp-json\/wp\/v2\/wiki\/477882\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/oneproxy.pro\/cn\/wp-json\/wp\/v2\/media\/468808"}],"wp:attachment":[{"href":"https:\/\/oneproxy.pro\/cn\/wp-json\/wp\/v2\/media?parent=477882"}],"curies":[{"name":"\u53ef\u6e7f\u6027\u7c89\u5242","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}