ã§ãã‚‹ã ã‘ä¸å¯§ã«GANã¨DCGANã‚’ç†è§£ã™ã‚‹ - åˆç¡äºŒæ™‚å››å分目的 Chainerã®æ‰±ã„ã«æ…£ã‚Œã¦ããŸã®ã§ã€ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã‚’使ã£ãŸç”»åƒç”Ÿæˆã«æ‰‹ã‚’出ã—ã¦ã¿ãŸã„ ã„ã‚ã„ã‚ãªæ‰‹æ³•ãŒæ案ã•ã‚Œã¦ã„ã‚‹ãŒã€ã¾ãšã¯ä»Šå¹´å§‹ã‚ã«è©±é¡Œã«ãªã£ãŸDCGANを実際ã«è©¦ã—ã¦ã¿ã‚‹ãŸã„ ãã®ãŸã‚ã«ã€ DCGANã‚’ã§ãã‚‹ã ã‘ä¸å¯§ã«ç†è§£ã™ã‚‹ã“ã¨ãŒã“ã®ã‚¨ãƒ³ãƒˆãƒªã®ç›®çš„ å°†æ¥GAN / DCGANを触る人ã®åŠ©ã‘ã«ãªã£ãŸã‚Šã€ç†è§£é–“é•ã£ã¦ã‚‹ã¨ã“ã‚ã«ãƒ„ッコミãŒã‚ã‚‹ã¨å¬‰ã—ㄠ本エントリã®æ§‹æˆ DCGANã®å‰æã¨ãªã£ã¦ã„ã‚‹GANã®è«–æ–‡ã®è¦ç‚¹ã‚’ã¾ã¨ã‚ã‚‹ DCGANã§GANã¨ã®å·®åˆ†ã¨ã—ã¦æ案ã•ã‚Œã¦ã„ã‚‹è¦ç‚¹ã‚’ã¾ã¨ã‚ã‚‹ DCGANã®mattyaã•ã‚“ã®å®Ÿè£…ã‚’èªã¿é€šã—ã¦è©³ç´°ã‚’ç†è§£ã™ã‚‹ 1. GANã«ã¤ã„㦠GANã¯ã€ã‚µãƒ³ãƒ—ル群ã¨ä¼¼ãŸã‚ˆã†ãªæ€§è³ªã‚’æŒã¤å‡ºåŠ›ã‚’生æˆã™ã‚‹ãŸã‚ã®ãƒ•ãƒ¬ãƒ¼ãƒ ワーク 2014å¹´ã«Ian J. Goodfellowらã«ã‚ˆã£ã¦æ案ã•ã‚ŒãŸ è«–æ–‡: Generative Adversarial Net
TensorFlowã«ã‚ˆã‚‹DCGANã§ã‚¢ã‚¤ãƒ‰ãƒ«ã®é¡”ç”»åƒç”Ÿæˆ - ã™ãŽã‚ƒãƒ¼ã‚“メモ
ã¯ã˜ã‚ã¦ã®GAN
{{#tags}}- {{label}}
{{/tags}}