30分ã§å®Œå…¨ç†è§£ã™ã‚‹Transformerã®ä¸–ç•Œã¯ã˜ã‚ã« åˆã‚ã¾ã—ã¦ã€‚ZENKIGENデータサイエンスãƒãƒ¼ãƒ ã®ã¯ã¾ãªã™ã§ã™ã€‚æ£å¼ãªæ‰€å±žã¯DeNAデータ本部AI技術開発部ãªã®ã§ã™ãŒ[1]ã€æ¥å‹™å§”託ã¨ã„ã†å½¢ã§ä»Šå¹´åº¦ã‹ã‚‰æ·±å±¤å¦ç¿’ç³»ã®é–‹ç™ºç‰ã«æºã‚ã£ã¦ã„ã¾ã™ã€‚ 深層å¦ç¿’界隈ã§ã¯ã€2017å¹´ã«è¡æ’ƒçš„ãªã‚¿ã‚¤ãƒˆãƒ«ï¼ˆAttention Is All You Need）ã®è«–æ–‡ãŒç™ºè¡¨ã•ã‚Œã¦ã‹ã‚‰æ—©5å¹´åŠã€å…ƒå‡ºè‡ªã®æ©Ÿæ¢°ç¿»è¨³ã‚¿ã‚¹ã‚¯ã‚’大ãã越ãˆã¦ã€Transformer関連ã®æŠ€è¡“ãŒæ§˜ã€…ãªé ˜åŸŸã§ç”¨ã„られる汎用アーã‚テクãƒãƒ£ã¨ã—ã¦ç›®è¦šã¾ã—ã発展ã—続ã‘ã¦ã„ã¾ã™ã€‚ 今回ã¯ãã‚“ãªTransformerãŒç¾æ™‚点ã¾ã§ã«ã©ã®ã‚ˆã†ã«æ´»ç”¨ã•ã‚Œã¦ããŸã‹ã€ã¾ãŸã€ã©ã®ã‚ˆã†ã«å·¥å¤«ã•ã‚Œã¦ããŸã‹ã‚’ã–ã£ãã‚Šã¨ä¿¯çž°ã—ã€æµã‚Œã‚’ãŠã•ã‚‰ã„ã™ã‚‹ç›®çš„ã®è¨˜äº‹ã«ãªã‚Šã¾ã™ã€‚本記事ã®å¤§æž ã¯ã€2021年時点ã§ã®ã‚µãƒ¼ãƒ™ã‚¤è«–æ–‡ã§ã‚ã‚‹ A Survey of Transformers ã«å€£ã„ã¤ã¤ã€é©å®œã€2023å¹´2月上旬ç¾åœ¨ã¾
Unlocking data analytics and machine learning for more businesses | Google Cloud Blog
Convolutional Neural Networkã«å¯¾ã™ã‚‹æ”»æ’ƒæ‰‹æ³• -誤分類ã®èª˜ç™º- | 技術者ブãƒã‚° | 三井物産セã‚ãƒ¥ã‚¢ãƒ‡ã‚£ãƒ¬ã‚¯ã‚·ãƒ§ãƒ³æ ªå¼ä¼šç¤¾
https://reinforce.io/blog/introduction-to-tensorforce/
Googleã€æ©Ÿæ¢°å¦ç¿’をスマートフォンã ã‘ã§å‡¦ç†ã™ã‚‹ä½Žé…延ã€ä½Žé›»åŠ›ã®ç”»åƒèªè˜æŠ€è¡“ã«ãŠã‘るコンピュータビジョンモデル「MobileNetsã€ã‚’GitHubã«ã¦å…¬é–‹
https://gradient.paperspace.com/
Kaggleã¾ã¨ã‚:BOSCH(kernels) - Qiita
自然言語処ç†ã«ãŠã‘る畳ã¿è¾¼ã¿ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ã‚’ç†è§£ã™ã‚‹ · ã‘ã‚“ã”ã®ãŠå±‹æ•·
代表的ãªæ©Ÿæ¢°å¦ç¿’手法一覧 - Qiita
https://github.com/sbrugman/deep-learning-papers/blob/master/README.md
Tuning Recurrent Neural Networks with Reinforcement Learning
IBMã®äººå·¥çŸ¥èƒ½ã€Œãƒ¯ãƒˆã‚½ãƒ³ã€ã€åŒ»è€…ãŒæ€ã„もよらã¬æ²»ç™‚法を続々発見（å°æž— 雅一） @gendai_biz
YouTubeã®ãƒ¬ã‚³ãƒ¡ãƒ³ãƒ‡ãƒ¼ã‚·ãƒ§ãƒ³ã‚¢ãƒ«ã‚´ãƒªã‚ºãƒ ã¯ã©ã®ã‚ˆã†ã«å‹•ãã‹
NeuroEvolution : Flappy Bird
機械å¦ç¿’ã‚’å¦ã¶ä¸Šã§æŠ‘ãˆã¦ãŠããŸã„æ•°å¦ï¼’ - HELLO CYBERNETICS
Googleã®ãƒ‹ãƒ¥ãƒ¼ãƒ©ãƒ«ãƒãƒƒãƒˆãƒ¯ãƒ¼ã‚¯ãŒè‡ªã‚‰æš—å·åŒ–通信を考案
MLDB: The Machine Learning Database - KDNuggets
MLDB Documentation
ç”»åƒèªè˜ã®æ€§èƒ½ã¯åœ§å€’çš„ã€ä»–é ˜åŸŸã§ã¯ã¾ã 発展途上
Announcing YouTube-8M: A Large and Diverse Labeled Video Dataset for Video Under
{{#tags}}- {{label}}
{{/tags}}