{"type":"rich","width":"100%","blog_title":"\u767a\u58f0\u7df4\u7fd2","author_name":"next49","categories":["\u30e1\u30e2"],"author_url":"https://blog.hatena.ne.jp/next49/","description":"Google DeepMind\u793e\u304c\u958b\u767a\u3057\u3066\u3044\u308bAlpha Go\u306e\u8ad6\u6587\u3092\u8aad\u3093\u3067\u307f\u305f\u3051\u3069\u3001\u3046\u3063\u3059\u3089\u3068\u3057\u304b\u7406\u89e3\u3067\u304d\u306a\u3044\u3002\u6a5f\u68b0\u5b66\u7fd2\u306e\u77e5\u8b58\u304c\u8db3\u308a\u306a\u3044\u3002stochastic gradient descent\u3068stochastic gradient ascent\u3063\u3066\u306a\u3093\u3060\u3088\uff01\u306a\u3093\u304b\u306e\u5fc5\u6bba\u6280\u306b\u3057\u304b\u898b\u3048\u306a\u3044\u3002\u3044\u308d\u3044\u308d\u8abf\u3079\u306a\u304c\u3089\u8aad\u3093\u3067\u307f\u308b\u3068\u3001\u30b3\u30f3\u30d4\u30e5\u30fc\u30bf\u56f2\u7881\u304c\u5287\u7684\u306b\u5f37\u304f\u306a\u3063\u305f\u539f\u56e0\u3067\u3042\u308b\u30e2\u30f3\u30c6\u30ab\u30eb\u30ed\u6728\u63a2\u7d22\u306b\u304a\u3051\u308b\u30d7\u30ec\u30a4\u30a2\u30a6\u30c8\uff08Nature\u306e\u8ad6\u6587\u3060\u3068rollout\uff09\u6642\u306e\u6253\u3061\u65b9\uff08Policy network\uff09\u3068\u30e2\u30f3\u30c6\u30ab\u30eb\u30ed\u6728\u63a2\u7d22\u306e\u30ce\u30fc\u30c9\u306e\u8a55\u4fa1\u5024\uff08value network\uff09\u3092\u6c7a\u3081\u308b\u3068\u3053\u308d\u306b\u7573\u307f\u8fbc\u307f\u30cb\u30e5\u30fc\u30e9\u30eb\u30cd\u30c3\u30c8\u30ef\u30fc\u30af\u2026","url":"https://next49.hatenadiary.jp/entry/20160315/1457970500","blog_url":"https://next49.hatenadiary.jp/","version":"1.0","title":" Alpha Go\u306e\u8a71","provider_url":"https://hatena.blog","provider_name":"Hatena Blog","height":"190","image_url":null,"published":"2016-03-15 00:48:20","html":""}