{"id":3819,"date":"2022-06-17T09:20:03","date_gmt":"2022-06-17T09:20:03","guid":{"rendered":"https:\/\/www.synsense.ai\/?p=3819"},"modified":"2023-03-13T07:10:38","modified_gmt":"2023-03-13T07:10:38","slug":"synsense-exodus-efficient-training-of-spiking-neural-networks","status":"publish","type":"post","link":"https:\/\/www.synsense.ai\/zh-hans\/synsense-exodus-efficient-training-of-spiking-neural-networks\/","title":{"rendered":"SynSense\u65f6\u8bc6\u79d1\u6280\u63d0\u51fa\u8109\u51b2\u795e\u7ecf\u7f51\u7edc\u8bad\u7ec3\u65b0\u65b9\u6cd5EXODUS\uff0c\u63d0\u5347\u8bad\u7ec3\u901f\u5ea6\u3001\u7a33\u5b9a\u6027\u53ca\u51c6\u786e\u6027 | \u6280\u672f\u7126\u70b9"},"content":{"rendered":"

\u8fd1\u65e5\uff0cSynSense\u65f6\u8bc6\u79d1\u6280<\/a>\u6280\u672f\u56e2\u961f\u53d1\u8868\u9898\u4e3a\u201cEXODUS: Stable and Efficient Training of Spiking Neural Networks\u201d\u7684\u6587\u7ae0\uff0c\u5728\u6587\u7ae0\u4e2d\u63d0\u51fa\u4e86\u65b0\u7684\u8109\u51b2\u795e\u7ecf\u7f51\u7edc\u8bad\u7ec3\u65b9\u6cd5EXODUS\u3002\u7814\u7a76\u663e\u793a\uff1aEXODUS\u80fd\u591f\u5927\u5927\u51cf\u4f4e\u8bad\u7ec3\u590d\u6742\u7a0b\u5ea6\uff0c\u66f4\u5177\u7a33\u5b9a\u6027\uff0c\u5e76\u63d0\u5347\u4e86\u51c6\u786e\u5ea6\u3002<\/p>\n

\u8fd9\u9879\u7814\u7a76\u7531SynSense\u65f6\u8bc6\u79d1\u6280\u745e\u58eb\u6280\u672f\u56e2\u961f\u7b97\u6cd5\u5e94\u7528\u3001\u673a\u5668\u5b66\u4e60\u5de5\u7a0b\u5e08\u5171\u540c\u5b8c\u6210\u3002
\n\u5177\u4f53\u89c1\uff1ahttps:\/\/arxiv.org\/abs\/2205.10242v1<\/a>
\nCode: https:\/\/github.com\/synsense\/sinabs-exodus<\/a><\/p>\n

\u201cEXODUS\u662f\u4e00\u79cd\u66f4\u52a0\u4e25\u8c28\u7684\u7b97\u6cd5\uff0c\u5bf9SLAYER\u8fdb\u884c\u4e86\u4f18\u5316\uff0c\u5728\u591a\u79cd\u57fa\u51c6\u6570\u636e\u96c6\u4e2d\u663e\u793a\u51fa\u4e86\u66f4\u5feb\u901f\u7684\u6536\u655b\u548c\u66f4\u9ad8\u7684\u51c6\u786e\u6027\u3002\u201d<\/p>\n

\u2014\u2014SynSense\u65f6\u8bc6\u79d1\u6280<\/p>\n

\u9ad8\u7ea7\u7b97\u6cd5\u603b\u76d1Sadique Sheik<\/p>\n

\"\"<\/p>\n

\u5728\u5982\u4eca\u7684\u673a\u5668\u5b66\u4e60\u4efb\u52a1\u4e2d\uff0c\u529f\u8017\u6b63\u5728\u88ab\u7f6e\u4e8e\u8d8a\u6765\u8d8a\u91cd\u8981\u7684\u5730\u4f4d\u3002\u53d7\u751f\u7269\u5927\u8111\u542f\u53d1\u7684\u8109\u51b2\u795e\u7ecf\u7f51\u7edc\u5c24\u5176\u5728\u8fde\u63a5\u4e8b\u4ef6\u9a71\u52a8\u7684\u4f20\u611f\u5668\u53ca\u5f02\u6b65\u786c\u4ef6\u65f6\uff0c\u663e\u793a\u51fa\u7a81\u51fa\u7684\u4f4e\u529f\u8017\u3001\u4f4e\u5ef6\u8fdf\u7279\u5f81\u3002\u56e0\u6b64\uff0c\u8d85\u4f4e\u529f\u8017\u8109\u51b2\u795e\u7ecf\u7f51\u7edc\u4e5f\u6b63\u5728\u88ab\u5e7f\u6cdb\u63a5\u53d7\u3002<\/p>\n

\u7814\u7a76\u4eba\u5458\u8bbe\u8ba1\u4e86\u4e0d\u540c\u7684\u65b9\u6cd5\u7528\u4e8e\u8109\u51b2\u795e\u7ecf\u7f51\u7edc\u8bad\u7ec3\u3002\u5f53\u524d\uff0c\u8bad\u7ec3\u8109\u51b2\u795e\u7ecf\u7f51\u7edc\u6240\u8fd0\u7528\u7684\u57fa\u4e8e\u65f6\u95f4\u7684\u53cd\u5411\u4f20\u64ad\uff08BPTT\uff09\uff0c\u5374\u8f83\u4e3a\u8017\u65f6\u3002\u6765\u81eaNUS\u7684Shrestha\u548cOrchard\u6b64\u524d\u63d0\u51fa\u4e86\u4e00\u79cd\u540d\u4e3aSLAYER\u7684\u7b97\u6cd5\uff0c\u80fd\u591f\u76f8\u5f53\u7a0b\u5ea6\u4e0a\u63d0\u5347\u8bad\u7ec3\u901f\u5ea6\u3002\u7136\u800c\uff0cSLAYER\u5e76\u672a\u5728\u8ba1\u7b97\u68af\u5ea6\u4e2d\u5c06\u795e\u7ecf\u5143\u819c\u7535\u4f4d\u91cd\u7f6e\u673a\u5236\u7eb3\u5165\u8003\u91cf\uff0c\u8fd9\u4e5f\u662f\u9020\u6210\u6570\u503c\u4e0d\u7a33\u5b9a\u7684\u4e00\u5927\u539f\u56e0\u3002\u4e3a\u4e86\u5bf9\u6297\u8fd9\u4e00\u7f3a\u9677\uff0cSLAYER\u5f15\u5165\u4e86\u68af\u5ea6\u7f29\u653e\u8d85\u53c2\u6570\uff0c\u5e76\u4e14\u9700\u8981\u624b\u52a8\u8c03\u8bd5\u3002<\/p>\n

SynSense\u65f6\u8bc6\u79d1\u6280\u5728\u6700\u65b0\u7814\u7a76\u4e2d\uff0c\u5bf9SLAYER\u8fdb\u884c\u4e86\u6539\u5584\u5e76\u8bbe\u8ba1\u4e86\u4e00\u79cd\u540d\u4e3aEXODUS\u7684\u7b97\u6cd5\u3002EXODUS\u5c06\u795e\u7ecf\u5143\u819c\u7535\u4f4d\u91cd\u7f6e\u673a\u5236\u8003\u8651\u5728\u5185\uff0c\u5e94\u7528\u4e86\u9690\u51fd\u6570\u5b9a\u7406\uff08 Implicit Function Theorem \/IFT\uff09 \u6765\u8ba1\u7b97\u68af\u5ea6\uff0c\u65e0\u9700\u7279\u7f6e\u68af\u5ea6\u7f29\u653e\uff0c\u5927\u5927\u51cf\u4f4e\u4e86\u8bad\u7ec3\u590d\u6742\u7a0b\u5ea6\u3002<\/p>\n

\"\"
\nSynSense\u65f6\u8bc6\u79d1\u6280\u5728\u6b64\u7814\u7a76\u8fc7\u7a0b\u4e2d\u6f14\u793a\u4e863\u7c7b\u7c7b\u8111\u4efb\u52a1\uff1a\u901a\u8fc7DVS Gesture\u3001Heidelberg Spiking Digits\uff08HSD)\u3001Spiking speech commands \uff08SSC) \uff0c\u56f4\u7ed5EXODUS\u53caSLAYER\u8fdb\u884c\u4e86\u5bf9\u6bd4\uff0cEXODUS\u663e\u793a\u51fa\u4e86\u8f83SLAYER\u66f4\u9ad8\u7684\u51c6\u786e\u6027\u3002<\/p>\n

\"\"
\n\u9a8c\u8bc1\u51c6\u786e\u5ea6\u63d0\u5347\u53ca\u8bad\u7ec3\u63d0\u901f\u5bf9\u6bd4<\/p>\n

\u81ea\u6210\u7acb\u4ee5\u6765\uff0cSynSense\u65f6\u8bc6\u79d1\u6280\u6301\u7eed\u52a9\u63a8\u7c7b\u8111\u667a\u80fd\u4ea7\u4e1a\u843d\u5730\uff0c\u540c\u65f6\u5728\u8fd9\u4e00\u8fdb\u7a0b\u4e2d\uff0c\u4e8e\u8111\u79d1\u5b66\u3001\u795e\u7ecf\u7f51\u7edc\u6a21\u578b\u3001\u7b97\u6cd5\u3001\u786c\u4ef6\u67b6\u6784\u3001\u82af\u7247\u8bbe\u8ba1\u3001\u7cfb\u7edf\u7ea7\u8bbe\u8ba1\u3001\u5e94\u7528\u3001\u8f6f\u4ef6\u7b49\u7c7b\u8111\u5de5\u7a0b\u76f8\u5173\u7684\u5404\u4e2a\u9886\u57df\u6301\u7eed\u521b\u65b0\uff0c\u5bf9\u56fd\u5185\u5916\u79d1\u7814\u9886\u57df\u4ea7\u751f\u4e86\u79ef\u6781\u6df1\u523b\u7684\u5f71\u54cd\u3002SynSense\u65f6\u8bc6\u79d1\u6280\u6838\u5fc3\u56e2\u961f\u5728\u7c7b\u8111\u82af\u7247\u9886\u57df\u79d1\u7814\u4e0a\u7684\u5b9e\u529b\u4f4d\u4e8e\u4e16\u754c\u6700\u9876\u7ea7\u7684\u884c\u5217\uff0c\u8fd9\u4e5f\u5c06\u4f5c\u4e3a\u5f3a\u5927\u9a71\u52a8\uff0c\u52a0\u5febSynSense\u65f6\u8bc6\u79d1\u6280\u7c7b\u8111\u667a\u80fd\u5e94\u7528\u843d\u5730\uff0c\u4e3a\u7c7b\u8111\u6280\u672f\u7684\u66f4\u5e7f\u6cdb\u94fa\u5f00\u800c\u5f62\u6210\u52a9\u529b\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"

\u8fd1\u65e5\uff0cSynSense\u65f6\u8bc6\u79d1\u6280\u6280\u672f\u56e2\u961f\u53d1\u8868\u9898\u4e3a\u201cEXODUS: Stable and Efficient Tr<\/p>\n","protected":false},"author":10,"featured_media":3822,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"content-type":"","_mi_skip_tracking":false},"categories":[46],"tags":[],"acf":[],"_links":{"self":[{"href":"https:\/\/www.synsense.ai\/zh-hans\/wp-json\/wp\/v2\/posts\/3819"}],"collection":[{"href":"https:\/\/www.synsense.ai\/zh-hans\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.synsense.ai\/zh-hans\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.synsense.ai\/zh-hans\/wp-json\/wp\/v2\/users\/10"}],"replies":[{"embeddable":true,"href":"https:\/\/www.synsense.ai\/zh-hans\/wp-json\/wp\/v2\/comments?post=3819"}],"version-history":[{"count":3,"href":"https:\/\/www.synsense.ai\/zh-hans\/wp-json\/wp\/v2\/posts\/3819\/revisions"}],"predecessor-version":[{"id":4483,"href":"https:\/\/www.synsense.ai\/zh-hans\/wp-json\/wp\/v2\/posts\/3819\/revisions\/4483"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.synsense.ai\/zh-hans\/wp-json\/wp\/v2\/media\/3822"}],"wp:attachment":[{"href":"https:\/\/www.synsense.ai\/zh-hans\/wp-json\/wp\/v2\/media?parent=3819"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.synsense.ai\/zh-hans\/wp-json\/wp\/v2\/categories?post=3819"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.synsense.ai\/zh-hans\/wp-json\/wp\/v2\/tags?post=3819"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}