This domain may be for sale!
This domain may be for sale!
Deleted articles cannot be recovered. Draft of this article would be also deleted. Are you sure you want to delete this article? ããã¯Fujitsu Advent Calendar 2017ã®18æ¥ç®ã®è¨äºã§ãã æ²è¼å 容ã¯å¯å£«éã°ã«ã¼ãã代表ãããã®ã§ã¯ããã¾ããããã ããããã¾ã§ã®åãçµã¿ãè©ä¾¡ããã¦ãå¯å£«éãQiitaã«æ£å¼åå ãããã¨ã«ãªãã¾ãã[ãªã³ã¯]ããªããå 容ã®æ£ç¢ºæ§ã«ã¯æ³¨æãæã£ã¦ãã¾ãããç¡ä¿è¨¼ã§ãã ã¯ããã« ãã®è¨äºã§ã¯ä»å¹´çºè¡¨ããããã£ã¼ãã©ã¼ãã³ã°è«æï¼ArXivã§ã®çºè¡¨ææãçºè¡¨ãããå½éä¼è°ã2017å¹´éå¬ã¾ãã¯ã¸ã£ã¼ãã«æ²è¼ã2017å¹´ã®ãã®ï¼ããç§ãå人çã«éè¦ã ã¨æã£ãè«æãåéãã¦ãã¾ããã¾ãã2016å¹´æ«ããã®è«æãéè¦
2017å¹´12æã«éå¬ããããã¿ã¼ã³èªèã»ã¡ãã£ã¢ç解ç 究ä¼ï¼PRMUï¼ã«ã¦çºè¡¨ããç³ã¿è¾¼ã¿ãã¥ã¼ã©ã«ãããã¯ã¼ã¯ã®ãµã¼ã㤠ã2012å¹´ã®ç»åèªèã³ã³ããã£ã·ã§ã³ILSVRCã«ãããAlexNetã®ç»å ´ä»¥éï¼ç»åèªèã«ããã¦ã¯ç³ã¿è¾¼ã¿ãã¥ã¼ã©ã«ãããã¯ã¼ã¯ (CNN) ãç¨ãããã¨ãããã¡ã¯ãã¹ã¿ã³ãã¼ãã¨ãªã£ãï¼ILSVRCã§ã¯æ¯å¹´ã®ããã«æ°ããªCNNã®ã¢ãã«ãææ¡ããï¼ä¸è²«ãã¦èªè精度ã®åä¸ã«å¯ä¸ãã¦ããï¼CNNã¯ç»ååé¡ã ãã§ã¯ãªãï¼ã»ã°ã¡ã³ãã¼ã·ã§ã³ãç©ä½æ¤åºãªã©æ§ã ãªã¿ã¹ã¯ã解ãããã®ãã¼ã¹ãããã¯ã¼ã¯ã¨ãã¦ãåºãå©ç¨ããã¦ãã¦ããï¼ æ¬ç¨¿ã§ã¯ï¼AlexNet以éã®ä»£è¡¨çãªCNNã®å¤é·ãæ¯ãè¿ãã¨ã¨ãã«ï¼è¿å¹´ææ¡ããã¦ããæ§ã ãªCNNã®æ¹è¯ææ³ã«ã¤ãã¦ãµã¼ãã¤ãè¡ãï¼ããããå¹¾ã¤ãã®ã¢ããã¼ãã«åé¡ãï¼è§£èª¬ããï¼æ´ã«ï¼ä»£è¡¨çãªã¢ãã«ã«ã¤ãã¦è¤æ°ã®ãã¼ã¿ã»ãããç¨ãã¦å¦ç¿ããã³ç¶²
ãã®è¨äºã¯ç§»è»¢ãã¾ãããç´2ç§å¾ã«æ°è¨äºã¸ç§»åãã¾ãã移åããªãå ´åã¯ã³ã³ãã¯ãªãã¯ãã¦ãã ããã ãããªãã¨ããã¦ã¿ãã âãããããã pythonã«ããæ©æ¢°å¦ç¿ã®åå¼·ãããã®ã§ãå®è·µã¨ãããã¨ã§ã人æ°ã¢ã¤ãã«ãä¹æ¨å46ãã®å人çã«å¥½ããª5人ã®ã¡ã³ãã¼ãåºå¥ãã¦è¦ã¾ããã大ããªæµãã¯ãããªæãã§ãã webä¸ããäºäººã®ç»åã100æãã¤åã£ã¦ãã ç»åããé¡é¨åãåãåºãã¦ä¿åããã¹ããã¼ã¿ã®åãåºã ç»åã®æ°´å¢ã ã¢ãã«ãå®ç¾©ãã¦ãå¦ç¿ ãã¹ã(é¡ãåè§ãå²ã£ã¦ããã®äººã®ååãåºå) 説æã¯ãããªããã«ãã¦ã彼女ãã®å¯æãã«ã¤ãã¦èªãããã¨ããã§ãããããããããã°ã§ã¯ãªãã®ã§ãå°ãæè¡çãªãã¨ãæ¸ãã¾ãã ä»åã¯jupyterã使ã£ã¦ä½æ¥ãé²ãã¾ãããnotebookå½¢å¼ãªã®ã§çµæãè¦ãããåå¿è ã«ã¯ããç°å¢ã§ãããç°å¢ã¯ä»¥ä¸ã macOS:10.13.1 python:3.6.
6. 6 çµè·¯äºæ¸¬ã®æµã ⢠ç°å¢ããã®ç¹å¾´æ½åº â» ã·ã¼ã³ã©ãã« â» å±æé åã®ã³ã¹ã â» ã·ã¼ã³å ¨ä½ã®ç¹å¾´ãã¯ã㫠⢠対象ããã®ç¹å¾´æ½åº â» éå»ã®ç§»åçµè·¯ï¼åã ⻠身ä½çå±æ§ ⻠対象ã®ç¹å¾´ãã¯ãã« Prediction ç¹å¾´æ½åº äºæ¸¬ ⢠ãã¤ãºã¢ãã« â» KF, DBN ⢠ã¨ãã«ã®ã¼æå°å â» Dijkstraâs Algorithm ⢠深層å¦ç¿ â» CNN, LSTM ⢠éå¼·åå¦ç¿ ⢠ãã®ä» â» Social force model, Data driven å ¥å â» åç»å â» éå»ã®çµè·¯ åºå â» åº§æ¨ â» ç¢ºçåå¸ 7. 7 çµè·¯äºæ¸¬ã®æµã ⢠ç°å¢ããã®ç¹å¾´æ½åº â» ã·ã¼ã³ã©ãã« â» å±æé åã®ã³ã¹ã â» ã·ã¼ã³å ¨ä½ã®ç¹å¾´ãã¯ã㫠⢠対象ããã®ç¹å¾´æ½åº â» éå»ã®ç§»åçµè·¯ï¼åã ⻠身ä½çå±æ§ ⻠対象ã®ç¹å¾´ãã¯ãã« äºæ¸¬ Prediction ç¹å¾´æ½åº
This innocuous kitten photo, printed on a standard color printer, fools the classifier into thinking itâs a monitor or desktop computer regardless of how its zoomed or rotated. We expect further parameter tuning would also remove any human-visible artifacts. Out-of-the-box adversarial examples do fail under image transformations. Below, we show the same cat picture, adversarially perturbed to be i
çµè« ä¸è¨4ã¤ããªã¹ã¹ã¡ä¸ããé ã«è¦ã¦ããäºããªã¹ã¹ã¡ãã¾ãã ã4æ¥ã§ä½é¨ã TensorFlow x Python 3 ã§å¦ã¶ãã£ã¼ãã©ã¼ãã³ã°å ¥é ã¿ããªã®AIè¬åº§ ã¼ãããPythonã§å¦ã¶äººå·¥ç¥è½ã¨æ©æ¢°å¦ç¿ ã¼ãããä½ããã¥ã¼ã©ã«ãããã¯ã¼ã¯ãPython 3 + NumPyã§ããã¯ãããããå¾¹åºãã¹ã¿ã¼ã ã¢ããªã±ã¼ã·ã§ã³éçºè ã®ããã®æ©æ¢°å¦ç¿å®è·µè¬åº§ 大åæ ãã®è¨äºãæ¸ãã¦ã人 æ¬è· Railsãµã¼ãã¼ã¨ã³ã¸ãã¢ï¼iphoneã¢ã㪠ããã°ã©ãã§ãããï¼ ä¸ååã°ãä¸åæ³éãã¦ããããã°ã©ãã«ãªã£ãã å¤ãã種ã¨ãçªç¶å¤ç°ã¨è¨ããã¦ãã ä»å¾ã¯æ¯è¼ççæéï¼ä¸å¹´ä½ï¼ã§ãã¼ã¿ãµã¤ã¨ã³ãã£ã¹ãã«ãªã ãããªã給æã¢ãããä¼ãã§ããã æ³å®ãã¦ããèªè æ©æ¢°å¦ç¿ã ãã§ãªãããã°ã©ãã³ã°èªä½ãåå¿è ã«åãã¦ãã¾ãã åç½®ã Udemyã¨åç»å¦ç¿ã«ã¤ã㦠Udemyã¨ã¯ï¼ Udemy
ããã«ã¡ã¯ãã¹ãã¼ããã¥ã¼ã¹ã®å¾³æ°¸ã§ãã深層å¦ç¿æ¥çã¯GANã ã¨ã深層強åå¦ç¿ã ã¨ãã§çãä¸ãã£ã¦ãã¾ãããä»æ¥ã¯æ·¡ã ã¨ãã¹ãã¼ã¹ãªãã¥ã¼ã©ã«ãããã¯ã¼ã¯ã®è©±ããã¾ãã è¦ç´ãã㨠ãã¥ã¼ã©ã«ãããã®ã¹ãã¼ã¹åã«ãã£ã¦ã精度ã¯ã»ã¨ãã©ç ç²ã«ããã«ãè¨ç®å¹çã¯3ã5åç¨åº¦ã¾ã§åä¸ã§ãã ã¹ãã¼ã¹åã«ã¯ã¾ã 課é¡ãå¤ãããã¥ã¼ã©ã«ãããã®é«éåã¨ããæå³ã§ã¯ã次ã®æ¦å ´ã¯ãã¶ãããã«ãªã ã¹ãã¼ã¹ã¨ã¯ãã¹ãã¼ã¹åã¨ã¯ ã¹ãã¼ã¹ã§ããã¨ã¯ãå¤ã®ã»ã¨ãã©ã0ã§ãããã¨ã§ããä¾ãã°ããã¯ãã«$a,b$ã®å ç©ãè¨ç®ããéã«ã$a$ã®ã»ã¨ãã©ã®è¦ç´ ã®å¤ã0ã§ããã¨ãã¾ãããã0ã«ãªã«ãããã¦ã0ã§ãããã$a$ã®å¤ã0ã§ãªã次å ã«ã¤ãã¦ã®ã¿ã$a_i b_i$ã®å¤ãè¨ç®ãã¦è¶³ãåãããã°ããããã§ãããã®ããã«ãå ç©ãè¨ç®ããéã«ãã©ã¡ããã®ãã¯ãã«ãã¹ãã¼ã¹ã§ããã°è¨ç®ãé«éåã§ãã¾ãã0ã¨ããå¤ãã¡ã¢ãª
ã¯ããã¾ãã¦ãABEJAã§Researcherããããã¦ããã ãã¦ããç½å·ã§ãã å æ¥ãååç©ã®ç©æ§æ¨å®ãDeep Learningãã¤ãã£ã¦å¾æ¥ææ³ãã300,000åé«éã«å¦çããã¨ããè«æãã§ã¾ããï¼[1], [2]ï¼ããã®è«æã®ææ³ã¯ãGraph Convolutionã¨ããã°ã©ãä¸ã«å®ç¾©ãããConvolutionæ¼ç®ããã¼ã¹ã¨ãªã£ã¦ãã¾ããç©æ§æ¨å®ã«éãããã°ã©ã解æå ¨è¬ã Deep Learning ã§ä¸æã«ããªããããã«ãªãã°ãDeep Learningã®ã¢ããªã±ã¼ã·ã§ã³ã®å¹ ããã£ã¨æ¡ããããããªãã¤ããã¼ã·ã§ã³ãèµ·ããããªäºæããã¾ãã ICMLãNIPSãªã©ã®æ©æ¢°å¦ç¿ç³»ã®ä¸»è¦å½éä¼è°ã§ãæ°å¹´åããGraph Convolutionã«ã¤ãã¦ã®è«æãã¡ãã»ãåºç¾ãã¯ããã¦ãããã¨ãã«æè¿ãã®å¢ããå¢ãã¦ãã¦ããå°è±¡ãããã¾ããå人çã«ãæè¿ï¼åããï¼ï¼ã«ããã«ã°ã©ãã¥ãã¦ã
ãã¥ã¼ã©ã«ãããã¯ã¼ã¯ãç¨ããæ©æ¢°å¦ç¿ã®ããã£ã¼ãã©ã¼ãã³ã°(深層å¦ç¿)ãã¯äººå·¥ç¥è½(AI)éçºã«æ¬ ãããªãæè¡ã§ãããAI以å¤ã«ãGoogleãã©ãã®ç»åã®èªåã¿ã°ä»ãæ©è½ãAmazonã®ã¬ã³ã¡ã³ãæ©è½ãªã©ããã§ã«å®ç¨åããã¦ããæè¡ã«æ´»ç¨ããã¦ãã¾ãããããªãã£ã¼ãã©ã¼ãã³ã°ãéçºè ãå¦ç¿ããããã®ã¨ã£ã¦ããã®æ¹æ³ãGoogleã®ã¯ã©ã¦ãéçºè ãããã°ã§ç´¹ä»ãã¦ãã¾ãã Learn TensorFlow and deep learning, without a Ph.D. | Google Cloud Big Data and Machine Learning Blog  | Google Cloud Platform https://cloud.google.com/blog/big-data/2017/01/learn-tensorflow-and-deep-learning-w
åç½®ã# è¨äºãã¯ã¦ã¶çä¸ãã¦æ¥ããããæããããã®ã§ãçµæ§æ¸ãç´ãã¾ããã ãã®è¨äºãããè¯è³ªãªè¨äºãåèè¨äºã«åæããã®ã§ããã®ãã¼ã¸ãããã¯ãã¼ã¯éã ã¨ãã¦ãä»ã®ãã¼ã¸ãåç §ãã¦ããã ããã°ã¨æãã¾ãã ã¯ããã«# æ©æ¢°å¦ç¿ãåå¼·ããã«ããã£ã¦ã ãã¼ã¹ã¨ãªãæ°å¦ãåå¼·ãããã¨ããã¢ããã¼ã·ã§ã³ãé«ã¾ã£ã¦ããããªããï¼ããã¯ãä»ã¾ã§æ°å¦çãªç¥èãªãã«åå¼·ãé²ãã¦ããã®ãããã©ããè«æãèªããªãã£ãããå°ãæ°å¼ã§è¾¼ã¿å ¥ã£ã¦ããã¨ãã¨ããã«ãããããããªããªã£ãããã ã ããããä¸çªã®ã¢ããã¼ã·ã§ã³ã¯ããã£ã±ãæ©æ¢°å¦ç¿ãåå¼·ãããã®ã¨ãã¦ã®ç»ç«éãPRML(ãã¿ã¼ã³èªèã¨æ©æ¢°å¦ç¿)ãèªã¿ããã¨ãããã®ãããã åèè¨äº# ããã§ãæ©æ¢°å¦ç¿ã®ããã«å¿ è¦ãªæ°å¦ã調ã¹ã¦ã¿ãã®ã ãã»ã»ã»ã©ã®ãµã¤ããã¿ã¦ãããã¯ã¨ãããã®ãã¿ã¤ãããªãã®ã ã 2017å¹´ç¾å¨ã§ãæçãªè¨äºãã§ãããããéãã¦ã¿ãã
æ¦è¦ å æ¥çºè¡¨ãããpix2pixã楽ãããã ã£ãã®ã§å®éã«åããã¦ã¿ãã 対象ãçµãã°å¦ç¿ç»åã500æç¨åº¦ã§ããããªãã«åãã çµè«ï¼GANãããã ã¯ããã« pix2pixã¨ã¯ããã£ããã¨è¨ãã°ã2ç»åéã«æ½ãç»åå¤æãDNNã§è¡¨ç¾ãã¦ãã¾ããï¼ã¨ãããã®ã§ããå¤æåå¾ã®ç»åããç¨æã§ããã°å¤§ä½ã©ããªå¤æã§ã対å¿å¯è½ããããä¾ãã°ä»¥ä¸ã®ãããªãã¨ãå¯è½ãããã§ãã èªç©ºåç â å°å³ç»å ã¢ãã¯ãç»å â ã«ã©ã¼ç»å ç·ç» â ã«ã©ã¼ç»å (ä¸è¨ãã¹ã¦éãå¯è½) ä½ããä¸è½ã£ã½ãé°å²æ°ï¼ç´ æ´ããããã¨ã«ã³ã¼ããå ¬éããã¦ããï¼torchã ãã©ï¼ï¼ãããèªåç»åã§ãç°¡åã«è©¦ãããï¼ ã¨ãããã¨ã§è©¦ãã¦ã¿ã¾ããã ãã¼ã¿ flickrããåéããã©ã¼ã¡ã³ç»å(1,000æ)ã使ç¨ãå¤å°ã®ã´ãã¯ç¡è¦ã§ãå ¬å¼ã®ã½ã¼ã¹ã³ã¼ãã«åããã¦ãåç»åãä¸å¿ã§ã¯ãããï¼ãªãµã¤ãºãã¦ãããå¦ç¿ç»åã¨è©ä¾¡ç»
Photo via Visual Hunt å°ãåã®ãã¨ã§ãããAlphaGoã¨ããå²ç¢ã®äººå·¥ç¥è½ããã°ã©ã ãã¤ã»ã»ãã«ä¹æ®µã«åå©ãããã¨ã§è©±é¡ã«ãªãã¾ããã*1 ã¾ããä¸é¨ã®ã²ã¼ã ã«ããã¦ãDQNï¼Deep Q-networkï¼ãã人éãããä¸æããã¬ã¤ããããã«ãªã£ãã¨ãããã¥ã¼ã¹ã話é¡ã«ãªã£ã¦ãã¾ãããã*2 ä»åã¯ãããã®äºä¾ã§ä½¿ããã¦ããã深層強åå¦ç¿ãã¨ããä»çµã¿ã使ã£ã¦ãFXã®ã·ã¹ãã ãã¬ã¼ããã§ããªããã¨æãã調ã¹ã¦ã¿ã¾ããã 注æï¼å¼·åå¦ç¿ãFXãåå¼·ãå§ããã°ãããªã®ã§ãè²ã ééã£ã¦ããç®æãããããããã¾ããããææããã ããã¨å¹¸ãã§ãã ä»åã®å 容 1.å¼·åå¦ç¿ã«ã¤ã㦠1-1.å¼·åå¦ç¿ 1-2.Reinforcement Learning: An Introduction (2nd Edition) 1-3.UCL Course on RL 1-4.å¼·åå¦ç¿ã«ã¤ã
Deleted articles cannot be recovered. Draft of this article would be also deleted. Are you sure you want to delete this article? ããããããèªåé転è»ãã¯ã¦ã¯å²ç¢ã»å°æ£ã¨ãã£ãã²ã¼ã ã¾ã§ãæ¨ä»å¤ãã®ãAIããä¸éãã«ãããã¦ãã¾ãã ãã®ä¸ã®ãã¼ã¯ã¼ãã¨ãã¦ããå¼·åå¦ç¿ãã¨ãããã®ãããã¾ããããããæå³ã§ã¯ãæ°ããæ©æ¢°å¦ç¿ã®ææ³ã®ä¸ã§æã注ç®ããã¦ãã(ããã¦èªå¼µããã¦ããã»ã»ã»)ææ³ã¨ããããããããã¾ããã ä»åã¯ãã®å¼·åå¦ç¿ã¨ããææ³ã«ã¤ãã¦ãåºç¤ããæè¿ç®è¦ã¾ãã精度ãåºãã¦ããDeep Q-learning(ããããããã¥ã³ãDQNã§ã)ã¾ã§ããã®çºå±ã®æµãã¨ä»çµã¿ã«ã¤ãã¦è§£èª¬ããã¦ããããã¨æãã¾ãã æ¬è¨äºã®å 容ããã¼ã¹ã«ããã³ãºãªã³ã¤ãã³ããé
3. æ©æ¢°å¦ç¿ã®ããã®è¡åã®å¾®å x =    x1 ... xk   , f(x) ã¯ã¹ã«ã©ã¼, F(x) =    f1(x) ... fm(x)   , a =    a1 ... ak    ã¨ãã. âF (x) âx =     âf1(x) âx1 · · · âfm(x) âx1 ... ... ... âf1(x) âxk · · · âfm(x) âxk    , âxT a âx = âaT x âx = a åå¾®åæ¼ç®åã§F(x)ãä½ç¨ãã¦ããã¨è§£é âTr(AB) âaij = bji â âTr(AB) âA = BT A,Bãè¡åã¨ãã. åæé¢æ°ã®å¾®å ãç´æéã âf(g(x)) âx = âg(x) âx âf(g(x)) âg(x) ä»é²ï¼ï¼æ°å¦ã®å¾©ç¿http://www.r.dl.itc.u-tok
ãPythonãã¼ã¿åæåå¼·ä¼#33ããDearPyGuiã«å ¥éãã¾ãããã®ç¶ãï½Image-Processing-Node-Editorï½
An Interactive Node-Link Visualization of Convolutional Neural Networks Abstract Convolutional neural networks are at the core of state-of-the-art approaches to a variety of computer vision tasks. Visualizations of neural networks typically take the form of static node-link diagrams, which illustrate only the structure of a network, rather than the behavior. Motivated by this observation, this pap
ã©ã³ãã³ã°
ã¡ã³ããã³ã¹
ãç¥ãã
ãªãªã¼ã¹ãé害æ å ±ãªã©ã®ãµã¼ãã¹ã®ãç¥ãã
ææ°ã®äººæ°ã¨ã³ããªã¼ã®é ä¿¡
å¦çãå®è¡ä¸ã§ã
j次ã®ããã¯ãã¼ã¯
kåã®ããã¯ãã¼ã¯
lãã¨ã§èªã
eã³ã¡ã³ãä¸è¦§ãéã
oãã¼ã¸ãéã
{{#tags}}- {{label}}
{{/tags}}