Notice: file_put_contents(): Write of 30 bytes failed with errno=28 No space left on device in /var/www/html/uzcms/firstvoltllc.com/incs/data.php on line 1473 Notice: fwrite(): Write of 32 bytes failed with errno=28 No space left on device in /var/www/html/uzcms/firstvoltllc.com/incs/data.php on line 113 Notice: fwrite(): Write of 1827 bytes failed with errno=28 No space left on device in /var/www/html/uzcms/firstvoltllc.com/incs/data.php on line 113 {"id":2919,"date":"2022-06-22T15:42:00","date_gmt":"2022-06-22T06:42:00","guid":{"rendered":"https:\/\/kakaoenterprise.com\/?post_type=about&p=2919"},"modified":"2022-07-27T14:15:51","modified_gmt":"2022-07-27T05:15:51","slug":"%ec%b9%b4%ec%b9%b4%ec%98%a4%ec%97%94%ed%84%b0%ed%94%84%eb%9d%bc%ec%9d%b4%ec%a6%88-ai%e3%83%bb%eb%8d%b0%ec%9d%b4%ed%84%b0%eb%a7%88%ec%9d%b4%eb%8b%9d-%eb%b6%84%ec%95%bc-%ea%b5%ad%ec%a0%9c-%ed%95%99","status":"publish","type":"about","link":"https:\/\/kakaoenterprise.com\/about\/%ec%b9%b4%ec%b9%b4%ec%98%a4%ec%97%94%ed%84%b0%ed%94%84%eb%9d%bc%ec%9d%b4%ec%a6%88-ai%e3%83%bb%eb%8d%b0%ec%9d%b4%ed%84%b0%eb%a7%88%ec%9d%b4%eb%8b%9d-%eb%b6%84%ec%95%bc-%ea%b5%ad%ec%a0%9c-%ed%95%99\/","title":{"rendered":"AI\u30fb\ub370\uc774\ud130\ub9c8\uc774\ub2dd \ubd84\uc57c \uad6d\uc81c \ud559\ud68c SIGKDD\uc5d0\uc11c \uc5f0\uad6c \uc131\uacfc \ubc1c\ud45c"},"content":{"rendered":"\n

1) Connecting a Low Loss Subspace for Personalized Federated Learning<\/p>\n\n\n\n

\uc5f0\ud569 \ud559\uc2b5(Federated Learning)\uc740 \ud074\ub77c\uc774\uc5b8\ud2b8 \ub0b4\uc5d0 \uc800\uc7a5\ub41c \uc0ac\uc6a9\uc790\uc758 \ub370\uc774\ud130\ub97c \ubcf4\ub0b4\uc9c0 \uc54a\uace0, \ud074\ub77c\uc774\uc5b8\ud2b8\uc5d0\uc11c \uc77c\ubd80 \ud559\uc2b5\ub41c \ubaa8\ub378 \ud30c\ub77c\ubbf8\ud130\ub97c \uc911\uc559 \uc11c\ubc84\ub85c \uc9d1\uacc4\ud558\uc5ec \ud504\ub77c\uc774\ubc84\uc2dc \uc774\uc288\ub97c \ud574\uacb0\ud558\ub294 \ucd5c\uc2e0 \uba38\uc2e0\ub7ec\ub2dd \ubc29\ubc95\ub860\uc785\ub2c8\ub2e4. \ubcf8 \ub17c\ubb38\uc5d0\uc11c\ub294 \uac01 \ud074\ub77c\uc774\uc5b8\ud2b8 \ub610\ub294 \uac01 \uc0ac\uc6a9\uc790\uc5d0 \uac1c\uc778\ud654\ub41c \uc131\ub2a5\uc744 \ucd5c\ub300\ub85c \ubcf4\uc7a5\ud558\uae30 \uc704\ud574, \uc5f0\uacb0\uc131(Connectivity) \uc774\ub860\uc5d0 \uae30\ubc18\ud55c \uc0c8\ub85c\uc6b4 \uac1c\uc778\ud654 \uc5f0\ud569 \ud559\uc2b5 \uc54c\uace0\ub9ac\uc998\uc744 \uc81c\uc2dc\ud558\uc600\uc2b5\ub2c8\ub2e4. \uc774\ub97c \ud1b5\ud574 \uace0\uc131\ub2a5\uc758 \uac1c\uc778\ud654 \ubaa8\ub378\uc744 \ud6a8\uacfc\uc801\uc73c\ub85c \ud559\uc2b5\ud560 \uc218 \uc788\uc74c\uc744 \ubcf4\uc600\uace0, \ud5a5\ud6c4 \ud504\ub77c\uc774\ubc84\uc2dc\uac00 \uc911\uc694\ud55c \ub2e4\uc591\ud55c \uc751\uc6a9 \uc11c\ube44\uc2a4\uc5d0 \uc801\uc6a9\ub420 \uac83\uc73c\ub85c \uae30\ub300\ud569\ub2c8\ub2e4.<\/p>\n\n\n\n

? \uc790\uc138\ud788 \ubcf4\uae30<\/a><\/p>\n\n\n\n

2) X-ViT: High Performance Linear Vision Transformer without Softmax<\/p>\n\n\n\n

\uae30\uc874 ViT(Self-attention) \uc54c\uace0\ub9ac\uc998\uc774 \uac00\uc9c4 1) Token Size(Image Size)\uac00 \ub298\uc5b4\ub098\uba74 Model Complexity\uac00 O(N^2)\uc73c\ub85c \ucee4\uc9c0\ub294 \ubb38\uc81c\ub97c \uc5c6\uc560\uace0, 2) \uae30\uc874\uc5d0 \uc0ac\uc6a9\ud558\uc600\ub358 Softmax\ub97c L2 Normalization\uc73c\ub85c \ubcc0\uacbd\ud558\uc5ec Regularization \ud6a8\uacfc\ub97c \ub192\uc600\uc2b5\ub2c8\ub2e4. \uc774\ub97c \ud1a0\ub300\ub85c \ub9cc\ub4e0 X-ViT\ub294, \uae30\uc874 \uc54c\uace0\ub9ac\uc998\ubcf4\ub2e4 \ud6e8\uc52c \uacbd\ub7c9\ud654\ub41c \ud615\ud0dc\ub85c \uc131\ub2a5\uc744 \ub354 \ub192\uc774 \ub04c\uc5b4\uc62c\ub838\uc2b5\ub2c8\ub2e4.<\/p>\n\n\n\n

? \uc790\uc138\ud788 \ubcf4\uae30<\/a><\/p>\n\n\n\n

\uce74\uce74\uc624\uc5d4\ud130\ud504\ub77c\uc774\uc988\ub294 \uace0\ub3c4\ud654\ub41c AI \uae30\uc220 \uc5f0\uad6c\uc0ac\ub840\ub97c \uc9c0\uc18d\uc801\uc73c\ub85c \uacf5\uc720\ud558\uace0, \ub354\uc6b1 \ubc1c\uc804\ub41c \uc11c\ube44\uc2a4\uc758 \ubaa8\uc2b5\uc73c\ub85c \uc120\uc21c\ud658\uc2dc\ud0ac \uc608\uc815\uc785\ub2c8\ub2e4. \ub9ce\uc740 \uad00\uc2ec\uacfc \uc131\uc6d0 \ubd80\ud0c1\ub4dc\ub9bd\ub2c8\ub2e4.<\/p>\n","protected":false},"excerpt":{"rendered":"