2016-01-01ãã1å¹´éã®è¨äºä¸è¦§
çããããã«ã¡ã¯ ãå æ°ã§ããï¼ç§ã¯ã¼ã£ã¡ã§ããæ¬è¨äºã¯ãã«ãã¸ã§ã§ããªãæ©æ¢°å¦ç¿ã¨ã³ã¸ãã¢ãã®æçµæ¥ã§ãã qiita.comæ¬æ¥ã®è©±ã®æµãã¯æ¬¡ã®ã¨ããã§ãã ã¯ããã« ã«ãã¸ã§ãããªããã¨ã«å¯¾ããè§£æ±ºæ³ ã«ãã¸ã§ã«ã¤ãã¦ã®èå¯ ã«ãã¸ã§ã¨ã¯ çæ³ã®ã«â¦
çããããã«ã¡ã¯ ãå æ°ã§ããï¼å¹´æ«ã£ã¦ãããªã«å¿ãããªããã®ã§ãããæ¬è¨äºã¯Chainer Advent Calendar 16æ¥ç®ã®è¨äºã§ããqiita.comæ¬æ¥ã¯Scikit-learn likeãªChainerãä½ã£ãè¨äºã§ãããªããä½ã£ãã®ãã¯å¾è¿°ãã¾ãã ã¯ããã« ãªããæ½è±¡åã³ã¼ããâ¦
çããããã«ã¡ã¯ ãå æ°ã§ãããç§ã¯å æ°ã§ããæ¬è¨äºã¯Pythonã®ã¢ããã³ãã«ã¬ã³ãã¼ç¬¬6æ¥ã§ãã qiita.comæ¬æ¥ã¯Pythonã使ã£ãé³æ¥½è§£æã«ææ¦ãã¾ãã å¶ç¶ã«ãé³æ¥½è§£æã«ä¾¿å©ãªã©ã¤ãã©ãªãçºè¦ããã®ã§ã試ãã¦ã¿ããã¨æãã¾ãï¼ é³æ¥½è§£æ librosa liâ¦
çããããã«ã¡ã¯ ãå æ°ã§ãããã©ãã©ããæ¬è¨äºã¯DeepLearning Advent Calendar 3æ¥ç®ã®è¨äºã§ããqiita.comä»æ¥ã¯ç»åã«è²ãä»ä¸ãããããã¯ã¼ã¯ã«ã¤ãã¦çºè¡¨ãã¾ãã ãã£ãããªã®ã§ããã®ã¿ã¤ãã³ã°ãè¦è¨ãã£ã¦å ¬éãã¦ãã¾ãã çºè¡¨ã¹ã©ã¤ã æ ã¦ã¦â¦
çããããã«ã¡ã¯ ãå æ°ã§ãããæè¿ãChainer便å©ã§ã³ã£ãããããããªé åãã§ããé »ç¹ã«æ´æ°ãããã¨ã§æåãªChainerã§ãããä¹ ã ã«upgradeããã¨ä»¥åããã ã·ã³ãã«ãªã¿ã¹ã¯ã«ã¤ãã¦ãç°¡åã«å¦ç¿ãã§ãã¾ãã Trainer Chainer version 1.11.0ããTrainâ¦
çããããã«ã¡ã¯ ãå æ°ã§ããï¼ä¸äººã§ç®±æ ¹æ¸©æ³ããã£ããã§ãã¾ãããKeras.jsãåºã¦ãéã«Kerasãjavascriptãæ±ããããã«ãªãã¾ããã ï¼ããå ¬å¼ãªã®ãã©ãããé常ã«ä¸å®ã§ããã£ã¨éãï¼ ã¨ãããã¨ã§å®éã«åããã¦ã¿ããã¨æãã¾ãã Kerasã«ã¤ãâ¦
çããããã«ã¡ã¯ ãå æ°ã§ãããã©ã¼ã¡ã³é£ã¹ãããªã£ã¦ãããç§èªèº«ã¯èªä»å ±ã«èªããããªPython2.7.x userã®ã¤ãããªã®ã§ããã PyConJP2016ã§Andrey Vlasovskikhæ°ã®è¬æ¼ã§Type Hintsï¼åãã³ãï¼ã®è©±ããã£ã¦ Python3.5ãããã¨ãæããªããèãã¦ã¾ããâ¦
çããããã«ã¡ã¯ ãå æ°ã§ããï¼ç§ã¯å æ°ã§ããPythonã«ãããé«éåææ³ãæ²è¼ãã¦ã¿ã¾ãã ç°¡åãªã³ã¼ã並ã³ã«ç´¢å¼ã®ãããªæãã§å¼ãããããªã¤ã¡ã¼ã¸ã§ä½æãã¾ãããæ¬æ¥ã®ç®æ¬¡ã§ãã Pythonã«ãããé«éåã®å¿ è¦æ§ Pythonã®é«éå é«éåã®æé Profiâ¦
ã¿ãªããããã«ã¡ã¯ ãå æ°ã§ãããç§ã¯å æ°ã§ããJupyter Notebookã®æ¬¡ä¸ä»£çãJupyterLabãç´¹ä»ãããã¨æãã¾ããâ»7/17 誤åè±åãä¸é¨ç»åãä¿®æ£ JupyterLab JupyterLabã¨ã¯ JupyterLabã®ã¤ã³ã¹ãã¼ã« Jupyter Labã®åãç¹ ï¼ï¼ç»é¢åå²ãå¯è½ ï¼ï¼ã¿ãâ¦
çããããã«ã¡ã¯ ãå æ°ã§ãããç§ã¯å æ°ã§ããã¤ãã«ãä»åã¯dlibã§éãã§ã¿ã¾ããã â»dlibã¨ã¯ä½ã ï¼ã¨æãæ¹ã¯ä»¥ä¸ã®ãã¼ã¸ã¸nonbiri-tereka.hatenablog.com åºæ¬çãªæ§æ ç»åã®èªã¿è¾¼ã¿ã¨è¡¨ç¤º ç»åã®å¦çãããã¦ã¿ãã ç»åç¹å¾´ã«ä½¿ãããå¦çãå®æ½â¦
çããããã«ã¡ã¯ ãå æ°ã§ãããç§ã¯å æ°ã§ããæ¥æ¬ã§ã¯ãã¾ãè¦ãããªãdlibã¨å¼ã°ããã©ã¤ãã©ãªã®ç»åå¦çã©ã¤ãã©ãªã 使ã£ã¦ã¿ããã¨æãã¾ãã dlibã«ã¤ã㦠dlibã¨ã¯ å ¬å¼ãµã¤ãï¼dlib C++ Library Dlib is a modern C++ toolkit containing machineâ¦
çããããã«ã¡ã¯ ãå æ°ã§ãããç§ã¯å æ°ã§ãã æ¬æ¥ã¯Bandit Problemã¨å¼ã°ããåé¡ãå¼·åå¦ç¿ã§è§£ãã¦ã¿ã¾ãã Bandit Problemã«ã¤ã㦠解ãæ¹ ä»å解ããåé¡ epsilon greedy algorithm Softmax Tempature UCB ææ³ åèæç® ã½ã¼ã¹ã³ã¼ã Bandit Probleâ¦
çããããã«ã¡ã¯ ãå æ°ã§ãããç§ã¯å æ°ã§ããä»æ¥ã¯å¼·åå¦ç¿ã®åå¼·ãã¦ãã¡ã¢ãæ¸ãã¦ã¿ã¾ããã å人çã«ã¯æè¿ã注ç®ãã¦ããåéã§ãã´ã¼ã«ããããããªï¼ã¯ã©ã¹åé¡ãRegressionï¼ æ©æ¢°å¦ç¿ã¨ç°ãªããæ±ç¨çã«è²ã åé¡ã解ãããã ããã¨ããã®ãçç±â¦
çããããã«ã¡ã¯ ãå æ°ã§ãããååããããã¦ãã¾ãã¾ããããå¼ãç¶ã SensorBeeã®turtorialããã£ã¦ã¿ã¾ãããã¡ãã¯æºå段éã«2段éãããelasticsearch,kibanaã®ã¤ã³ã¹ãã¼ã«ã¨tutrialã®å®æ½ã§ãã æºå Elastisearch並ã³ã«kibanaã®ã¤ã³ã¹ãã¼ã« elasâ¦
æ¬æ¥ã¯ããã£ã¨SensorBeeã使ã£ã¦ã¿ã¾ãããçããããã«ã¡ã¯ ãå æ°ã§ãããé£ä¼ã§ã»ã£ã¨ãã¦ãã¾ãã What is SensorBee Stateful Expressive LightWeight Install Go(Mac) SensorBeeã®ã¤ã³ã¹ãã¼ã« Tutorial ãã¼ã¿ã®åå¾ã¨åæå ãµã¼ãã®èµ·å BQN ã¯ãâ¦
çããããã«ã¡ã¯ ãå æ°ã§ãããç§ã¯å æ°ã§ããä»æ¥ã¯ååã®ä»¥ä¸ã®ãã¼ã¸ããã ãã¶æ´æ°ãç«ã¡ãKaggleã®ã³ã³ãçã«ãå¤ãã®éå¬ãããã¾ããã ããã§ãæ°ãããã¼ã¸ã§ãªã³ã¯ãçºãã¦ã¿ã¾ããã ä¸ã«ã¯ã¤ã³ã¿ãã¥ã¼ããã©ã¼ã©ã ãgithubãªã©æ§ã ãªãã®ãæ··â¦
çããããã«ã¡ã¯ ãå æ°ã§ãããç§ã¯å æ°ã§ããä»æ¥ã¯çããNeural Networkã使ã£ã¦ããä¸ã§ã®å¤±æçµé¨ã«ã¤ãã¦èªãã¾ãã å¦ç¿ã®æã«æ¡å¤ã失æããã®ã§ãããã ãããåå ã¯æ±ºã¾ã£ã¦ãã¾ãããã大ä½ã¯ã»ã»ã» ã¨ãããã¨ã§ãä»åã¯å¤±æã®çµé¨ãã¢ã³ããã¿â¦
çããããã«ã¡ã¯ ãå æ°ã§ãããç§ã¯å æ°ã§ããå®ã¯ç§ãChainerã§ã®foræã§Linkã¨ãã¦ä½æã§ãããã¨ãç¥ããã ä»ã®è¤éãªãããã¯ã¼ã¯ã«Chainer使ãã«ããã¨æã£ã¦ãã¾ãããã以ä¸ã«ãµã³ãã«ããã£ã¦ ããããã°è¤éãªãããã¯ã¼ã¯ãçµãããã ãã¿ããâ¦
çããããã«ã¡ã¯ ãå æ°ã§ãããç§ã¯å æ°ã§ããä»æ¥ã¯Pythonã«ããã便å©è¨æ³ãæ¸ãã¦ã¿ã¾ãã ä»ã®è¨èªãã使ã£ããã¨ç¡ã人ããã²â»1/25ééããã¨æãããç®æã®å çä¿®æ£ãå®æ½ ãªã¹ãã«é¢ããã¤ãã¬ã¼ã·ã§ã³ enumerateï¼ã¤ã³ããã¯ã¹ã¨ãªãã¸ã§ã¯ãã®ã«â¦
çããããã«ã¡ã¯ ãå æ°ã§ãããç§ã¯å æ°ã§ããä»æ¥ã¯Pythonã«ãããã¡ã¢ãªå°ãªã使ãæ¹æ³ãç´¹ä»ãããã¨æãã¾ãã ãªãããããªæ¹æ³ãæ¸ãã«è³ã£ãããããã¯ããããªã¨ã©ã¼ãããã£ã¡ã ãè¦ã¦ããããã§ããã Traceback (most recent call last): File â¦
çããããã¾ãã¦ããã§ã¨ããããã¾ãã ãå æ°ã§ãããç§ã¯å æ°ã§ããæ°å¹´ã§ãã®ã§ä»å¹´ã®ãããããã¨ãæ¸ãã¦ã¿ããããªã¨æãã¾ãã Kaggle åå¼·ä¼ã§ã®çºè¡¨ èªæ¸ è±èª è«æ Kaggle æ¨å¹´åº¦ã¯æãã«æã¿ããªããªãBenchMarkã«ãåã¦ãªãç¶æ³ãç¶ãã¦ãããâ¦