AdaGradãŒ12å€é€Ÿããªã‚‹é”法AdaGradã¯å¦ç¿’率を自動調整ã—ã¦ãれる勾é…法ã®äºœç¨®ã§ã€ã„ã‚ã‚“ãªäººãŒçµ¶è³›ã—ã¦ã„ã¾ã™ã€‚ 勾é…を足ã—込む時ã«ã€å„次元ã”ã¨ã«ä»Šã¾ã§ã®å‹¾é…ã®2乗和をã¨ã£ã¦ãŠã„ã¦ã€ãã®å¹³æ–¹æ ¹ã§å‰²ã£ã¦ã‚ã’ã‚‹ã ã‘ã¨ã€æã‚ã—ãシンプルã§ã™ã€‚ Adaptive Subgradient Methods for Online Learning and Stochastic Optimization John Duchi, Elad Hazan, Yoram Singer. JMLR 2011. ä¸åº¦ã€ @echizen_tm ã•ã‚“ãŒãƒ–ãƒã‚°ã‚’書ã„ã¦ã¾ã—ãŸã€‚ AdaGrad+RDAを実装ã—ã¾ã—ãŸã€‚ 通常ã®SGDãªã©ã¯å¦ç¿’率をã ã‚“ã ん減衰ã•ã›ãªãŒã‚‰å‹¾é…を足ã—ã¦ã„ãã‚ã‘ã§ã™ãŒã€ã©ã®æ§˜ã«æ¸›è¡°ã•ã›ã‚‹ã‹ã¨ã„ã†å•é¡Œã«ã„ã¤ã‚‚é ã‚’æ‚©ã¾ã›ã¾ã™ã€‚ AdaGradã§ã¯æœ€åˆã®å¦ç¿’率ã“ã外ã‹ã‚‰ä¸Žãˆã¾ã™ãŒã€æ¸›è¡°ã®ã•ã›æ–¹ã‚„減衰率ã¨ã„ã£ãŸãƒã‚¤ãƒ‘ーパラメータã‹ã‚‰
ä¹±æŠžãƒ‡ãƒ¼ã‚¿æ§‹é€ ã®æœ€æ–°äº‹æƒ… −MinHash 㨠HyperLogLog ã®æœ€è¿‘ã®é€²æ©âˆ’ - iwiwiã®æ—¥è¨˜
{{#tags}}- {{label}}
{{/tags}}