Bayesian NMFを確率的変分ベイズ法ã§è§£ã - Qiita#ã¯ã˜ã‚ã« éžè² 値行列分解(Nonnegative Matrix Factorization;NMF)ã®ç¢ºçŽ‡çš„変分ベイズ法(stochastic variational Bayesian method)ã«ã‚ˆã‚‹è§£æ³•ã§ã™ã€‚ ã“ã¡ã‚‰ã®æ–‡çŒ®ã‚’ã¾ã¨ã‚ãŸã‚‚ã®ã§ã™ã€‚ #Bayesian NMF NMFã¯è¡Œåˆ—ã‚’éžè² 値ã®è¡Œåˆ—ã®ç©ã«åˆ†è§£ã™ã‚‹ã€æ¯Žåº¦ãŠãªã˜ã¿ã®ã‚ã‚Œã§ã™ã€‚分解後ã®è¡Œåˆ—ã®å„è¦ç´ ã«äº‹å‰åˆ†å¸ƒã‚’å°Žå…¥ã—ãŸã‚‚ã®ãŒBayesian NMFã§ã™ã€‚ åŒã˜ãƒ™ã‚¤ã‚ºçš„行列分解モデルã®LDAã¨ã®å¯¾å¿œã‚’考ãˆã‚‹ã¨ã€$\beta_{:k}$ã‚’topicã¨è¦‹ãªã›ã¾ã™(Bayesian NMFã®$X$ã¨LDAã®$W$ã¯è»¢ç½®ã®é–¢ä¿‚ã«ã‚ã‚‹ã“ã¨ã¯å°‘ã—注æ„ã§ã™)。 一方ã§$\sum_{v}\beta_{vk} \neq 1$ã¨ãªã‚‹ã¨ã“ã‚ãŒç•°ãªã‚Šã¾ã™ã€‚ #Stochastic variational Bayesian method 変分ベイズ法
éžè² å€¤è¡Œåˆ—å› å分解（NMF）ã«ã‚ˆã‚‹ãƒ¬ã‚³ãƒ¡ãƒ³ãƒ‰ã®ã¡ã‚‡ã£ã¨ã—ãŸä¾‹ - About connecting the dots.
{{#tags}}- {{label}}
{{/tags}}