t-SNE ã®å®Ÿè£…ã¯ã©ã‚Œã‚’使ã†ã¹ããªã®ã‹ï¼Ÿ - iwiwi 備忘録scikit-learn ã®å•é¡Œç‚¹ scikit-learn 信者ã¨ã—ã¦ã¯ã¨ã‚Šã‚ãˆãš scikit-learn ã®å®Ÿè£…を使ã„ãŸããªã‚‹ãŒã€scikit-learn ã®å®Ÿè£…ã¯ãŠã™ã™ã‚ã§ããªã„らã—ã„。 -https://www.red dit.com/r/MachineLearning/comments/47kf7w/scikitlearn_tsne_implementation/ （ã¯ã¦ãªãƒ–ãƒã‚°ã¯red ditã®URLを貼るã¨bad requestã«ãªã‚ŠæŠ•ç¨¿ã§ããªã„謎仕様） Besides being slower, sklearn's t-SNE implementation is fine once you realize the default learning rate is way too high for most applications. The definitive bh_t
How to Use t-SNE Effectively
{{#tags}}- {{label}}
{{/tags}}