è害ãšåŒã°ããããªã幎é ã§ããã2018幎3æã®1ã¶æéã§Courseraã§ã¹ã¿ã³ãã©ãŒã倧åŠãæäŸããŠããMachine Learningè¬åº§ãä¿®äºããŸãããå®å¹Žéè¿ã®èå µãã©ããã£ãŠãã®è¬åº§ãå®èµ°ããããããäœãç®æããŠããã®ããã話ãããããšæããŸãã
åæ©ïŒâ åæžïŒ
2æã®ãããµã2018ã§ããŠã«ã·ã¹ãã ãºã®æŒåããã®è¬æŒãèŽããŠãäžçãšã³ãžãã¢ã§é£¯ãé£ã£ãŠãããããªãäœãæ°ããããšãå§ããªãããã¡ã ãªãšæã£ãã®ãå§ãŸãã§ãã
2017幎ã¯ãXamarinã䜿ã£ãiOS/Androidã¢ããªãRaspbianäžã§C#ã䜿ã£ãREST APIãµãŒããŒãããã«ãiPhone/Apple Watchã«è©±ããããŠæ©æ¢°ãå¶åŸ¡ããVUIãªã©ãè£œé æ¥ã«å€ãããœãããšã³ãžãã¢ãšããŠã¯ããªãç°è³ªãªããšãããããŠããã£ããšæããŸãã
ã§ããæ°ããããšããã£ãŠãæãè¶³ãããäœãšãªãæ¶åäžè¯ãªæãã§ããã
ãæ°ããããšå§ãããâŠã
皌ãããšã³ãžãã¢ãç¶ããŠããããã«ããŸã ãŸã æ°ããæè¡ã«ææŠã§ããããšããèªä¿¡ã欲ãããã§ããã以äžã«ãäœãæ°ãããŠã¯ã¯ã¯ã¯ãããããªãã®ã«éŠãçªã£èŸŒã¿ãããšããè¡åã匷ããªã£ãŠããŸããã
ã¹ããã¯
詳现ã¯LinkedInãã芧ãã ãããããã«èšèŒãããŠããªãéšåããã£ãšã玹ä»ãããšãæ®éç§é«æ ¡ã忥ãã倧åŠã¯ååŠéšçµå¶åŠç§ã§äŒèšãåŠãã§ããçŽç²æç³»ã§ãåäŸã®é ããç®æ°ã¯èŠæã§ããè¶³ãç®ãšæãç®ã¯å€§äžå€«ã§ãããåŒãç®ãšå²ãç®ã奥ããã«ç¬ããããããèŠæã§ãã
ãããªåãçæ¶¯ã§äžåºŠã ããã¹ãã§çŸç¹ãåããŸãããããã埮ç©ã§ããèå µãªã®ã§é«æ ¡æ°åŠã§è¡åïŒæ°æå°èŠé ã§ã¯ãããããããããã§ããïŒãç¿ããŸãããäŒç€Ÿã§ã¯æµ·å€äºæ¥éšã«æå±ããŠãããTOEICã¯760ç¹ãããã§ãã客å ãšã®ãã£ã¹ã«ãã·ã§ã³ãªã©ã¯æèãã¯ã£ããããŠãã®ã§äžèªç±ãªãè¡ããŸããéã«é£äºã®åžãªã©ã§ã®éè«ã¯ãšãŠãé£ãããšããã¬ãã«ã§ãã
ç®ç
äŒç€Ÿã«å£²ãäžãããèªåã«åå ¥ãããšããæå³ã§çšŒãããšã³ãžãã¢ã§ããç¶ããããã§ããããã®ããã«ã¯èªåã®äŸ¡å€ãé«ããå¿ èŠããããŸããäžè¬çã«50ãéãããæ³ããšããã³ã«äŒç€Ÿå¡ã®äŸ¡å€ã¯äžãã£ãŠãããŸãããŸããäžè¬çãªäŒç€Ÿã®æè²äœç³»ã¯40代ãŸã§ã®äŒžã³ãããæåŸ ã§ããå±€ãã¿ãŒã²ããã«ããŠããŸãããããã£ãŠèå µã¯ãçµ¶ãã䟡å€ãåäžãããæœçãèªåèªèº«ã§èäœã«éæã£ãŠãããªããŠã¯ãããŸããã
ããèããæãäœãæã£ãŠäŸ¡å€ãåäžãããããšãããšãäžéã§ã¯è³ãã£ãŠããã©ç€Ÿå ã§ã¯åªã®ç¶æ ã§ããã°ãããã«çªã£èŸŒãã§ããããšã§å è¡è å©çã確ä¿ã§ããå¯èœæ§ãé«ããšæãããŸãã15幎åãããã£ãŠãã¢ãžã£ã€ã«éçºã10幎åŸã«å·šé¡ã®æ¡ä»¶ãåæ³šããåååã«ãªã£ãããã«ã
ããŸç€Ÿå ïŒæµ·å€å«ãïŒã«ã¯ãæ©æ¢°åŠç¿ã®çè«ããã¹ããã©ã¯ãã£ã¹ãäœç³»çã«åŠãã åºç€äœåã®ãã人æãããã®ãã©ããããåãããŸããããã ããã®ããŒããå€åœäººãšè±èªã§è©±ãåããæ¥æ¬äººãšã³ãžãã¢ãããªããšããã°ãããã¯åã«ãšã£ãŠå€§ããªãã£ã³ã¹ã§ããèŠããã«ãã®åéã¯å€åãŸã åªåªã®ãã«ãŒãªãŒã·ã£ã³ã§ãæ©ãè åã¡ã ã£ããããããããªãããšæã£ãŠããŸãã
äžæ¹ãä»äºãšããŠæ©æ¢°åŠç¿ã®ããŒãºã¯ããã®ãïŒãšãããšããããä»ã®ãšããè¯ãããããŸãããPoCã¬ãã«ã®æ¡ä»¶ãªã®ããèšèã ããäžäººæ©ãããŠããã®ããæ¬åœã«æ©æ¢°åŠç¿ãå¿ èŠãªæ¡ä»¶ãªã®ã粟æ»ãå¿ èŠãããããŸããããšã¯èšã£ãŠã粟æ»ã§ãã人æã¯ã©ãã«ïŒãšãã話ã§ããâŠ
ãšããããšã§ããã®è¬åº§ãä¿®äºãåºç€äœåãã€ããããŸã瀟å ã§ã¬ã¢ãªååšã«ãªããå¯èœæ§ã¯ãšãŠãé«ããã§ããããšã¯ãèªåããããæ©æ¢°åŠç¿ã§ãããïŒïŒéã声ïŒããšç©æ¥µçã«ã¢ããŒã«ããŠããã£ã³ã¹ãéãã¬ããã¢ã³ããã匵ãå·¡ããããã®æãæ¥ããïŒæ³£ããªããïŒäœ¿ããããã«ããã«ã¹ãã«ãæ·±ãåºããŠããããšã倧äºããšæãããŸããä»ã®ãšããã¯ã
èŠåŽããããš
ããŠãèå µã§æç³»è³ã§ãã®ã§è²ã ãšèŠåŽã¯ãããŸãããèçŒã§ããèŠããªããéäžåã2æéç¶ããªããå€èµ·ããŠããããªããšããè¥è ã«ã¯ç¡çžãªéå®³ãšæŠãããšããŸãã¯å¿ èŠã§ããããããããããã£ãé害ãèŠæ¹ãå€ãããšã¡ãªããã«ãªã£ããããã®ãé¢çœããšããã§ãã
äŸãã°ããå€èµ·ããŠããããªãããšããããšã¯è£ãè¿ããšãæèµ·ããã®ãæ©ãããšããããšã§ããåºæ¬çã«ã¯æ5æåã«èµ·ããŠ8æåã«ã¯äŒç€Ÿã«ãããšããå žåçãªãžãžã€ã®æŽ»åãã¿ãŒã³ã§ããããããé»è»ã¯åº§ããã空ããŠãã®ã§1æéã®é倿éã¯å匷ã«ã¡ããã©ãããiPhoneã§è¬çŸ©ã1æéèŠãã°ãã®é±ã®ååãããã®ãããªã¯çµãã£ãŠããŸããŸãã
äŒç€Ÿã«çããŠããå§æ¥ãŸã§ã®30ååã³ãæŒäŒã¿ã®45åã¯ãç¶ããã¿ããèŠçµãã£ããšããã®ããŒããã€ãããèªã¿è¿ããã埩ç¿ãããåºæ¬çã«å®æã§åž°ãã®ã§åž°ãã®é»è»ã1æéiPhoneã§è¬çŸ©ãèŠããããŒããèŠãŠåŸ©ç¿ãããééæéã ãã§ã3æéãæ¯æ¥äœ¿ããããã§ãã
ãããŠããã®ééæéãšããã®ãéåžžã«ããããã幎å¯ãã¯ãéäžåãç¶ããªããã®ã§ã3æéã¶ã£ç¶ãã«å匷ããããã30åçšåºŠã®çްåãæéã®æ¹ãéäžãä¿ã¡ãããã®ã§ããç¹ã«è¬çŸ©ã®ãããªã1æéãèŠç¶ãããåå以äžã¯äžã®ç©ºã§èãæµããŠããŸããŸããè±èªã§ãããã
ãŸããææžãã§ããŒããã€ãããšããã®ããããŸããããéäžåãç¶ããªãããšãšãé¢é£ããŸãããæãåãããŠåãæ°åŒãªã©ãæžããŠãããšéäžãç¶ãããããšãã广ããããŸããã·ã£ãã®ãã¿ã³ããšãã«ãããªã©çްããæäœæ¥ã«è¡°ããæãã仿¥ãã®é ããããã£ãæå ã䜿ã£ãŠåãæ°åŒãã°ã©ããªã©ãæžããŠèŠãããšããã®ããã±é²æ¢ã®è¯ããã¬ãŒãã³ã°ã«ãªããŸããäžç³äºé³¥ã§ãã
ãããããã°ã©ãã³ã°èª²é¡ã ãã¯å°ããŸãããããã°ããã¯ãŸãšãŸã£ãæéã確ä¿ããªããšã³ãŒããæžããŠããéäžã§ã¶ã£ãåãããããšã«ãªããæ¬¡ã«ãŸãæãåºããªããåéããã®ãã³ã³ããã¹ãã»ã¹ã€ããã»ã³ã¹ããé«ãã€ããŠéåžžã«å¹çãæªããããŸãã«é倿éã«MacBookéããŠã³ãŒãæžãããã«ããããªãããšããããšã§ãäž»ã«å€ãäŒæ¥ã«èªå® ã§ããŸãã¯æŒäŒã¿æéã«åãçµãã§ããŸãããæåã®2é±éã¯ã
åŸåã®2é±éã¯ãç¿å¹ŽåºŠããç°åïŒåºåå ããæ¬äœã«æ»ãïŒã®ãããœã£ãããšæéã空ããŠããŸããŸãããä»ãŸã§ã®ãããžã§ã¯ãã¯å®äºããŠã次ã®ãããžã§ã¯ãã¯æ°å¹ŽåºŠããã§ãªããšå§ãŸããªããšããããšã§ãåŸåã®2é±éãäžžã åŠç¿ã«ããŠãäžæ°ã«Machine Learningã®è¬åº§ãçµããŠããŸãããšã«ããŸããã
å ã ã¯ã5æã®GWåãããã«çµäºããããŒã¹ã§èšç»ããŠããã®ã§ãããäºæ³å€ã®ç©ºãæéãã§ããã®ã§ãããæå¹æŽ»çšãããŠããããŸãããå¥ã«éãã§ããããã§ããªããäžåžã®èš±å¯ãåŸãŠãŸãããåºåå ããæ¬äœã«æ»ããšããããšã§ãæ®ãæéãèªç±ã«ãããŠããã ããŸãããæè¬ã§ãã
èŠåŽããè¬åº§
ãã®è¬åº§ã®äžã§èŠåŽããããã°ã©ãã³ã°èª²é¡ã¯3ã€ãããŸãããŸã第äžã®é¢éã¯è¡åæäœãšOctaveãããåãããªãç¶æ ã§èšãã Week3ã®Logistic Regressionã2ã€ç®ã¯Week5ã®Backpropagationã3ã€ç®ãWeek7ã®Support Vector Machineã§ãããæåã®äºã€ã¯çè«ãšããŠã¯çè§£ã§ããã®ã§ãããããã°ã©ãã³ã°èª²é¡ããšãŠãé£ãããŠåã 5æéãããããã£ãŠããŸããŸããã
3ã€ç®ã®SVMã«è³ã£ãŠã¯æŠå¿µããããçè§£ã§ããäœåºŠããããªãèŠè¿ããŸããããæçµçã«ããããããšããšè ¹èœã¡ããã®ã¯ãã®ãµã€ãã®èª¬æã§ããããããŠããã°ã©ãã³ã°èª²é¡ã¯8æé以äžããã£ãŠ2æ¥ãããã§ã¯ãªã¢ãããšããæ ããªãã
ãã®ããã°ã©ãã³ã°èª²é¡ã®é£ãããšããã®ã¯ãåã«ãšã£ãŠã¯è¡åæäœã®é£ããã«å°œãããšèšã£ãŠãéèšã§ã¯ãããŸãããTransposeã ãšãInverseã ãšãElement-Wideã ãšããåžžã«è¡åã®æ¬¡å ãæèããªããé©ãã圢ã«å€ããŠããããšãæåã¯é£ããã£ãã§ãããã®ããã°ã©ãã³ã°èª²é¡ãã®ãã®ããè¡åæäœã«æ £ããããããã«äœãããŠããã®ã§ã¯ãªãããšåãããããªããããã«ã
ããã°ã©ãã³ã°èª²é¡ã¯ãããã«è§£çãèŒã£ãŠããã®ã§æ¢ãã°ãããã§ãã§ãŠããŸãããããã§ã¯èŠçŽéåã§ãããšã³ãžãã¢ãšããŠåè¬ããŠããæå³ããããŸãããã³ãŒã¹ã®ãµã€ãã«ã¯Resourcesãšããã¡ãã¥ãŒããã£ãŠãããã«Tutorialãšããã¡ã³ã¿ãŒãæžããŠããããã°ã©ã 仿§æžçãªèª¬æããããŸããããã«ã¯ã¹ãããã»ãã€ã»ã¹ãããã§ä»æ§ãèšè¿°ãããŠããŸãã®ã§ãè¡ãè©°ãŸã£ãããããèŠãããã«ããŸãããåãããªããããšãã£ãŠfor i=1:mã®ã«ãŒãã«éãããèŸããŠãè¡åæŒç®ã§ä¹ãåãããšãç®æããŸããã
ãŸãã¯ãèªåã§é 匵ã£ãŠèããŠã¿ããæ¥æ¬èªã®ãµã€ããå«ããŠåèæç®ã«ãåœãã£ãŠçè«ãçè§£ããã1åã«ã€ã1æéé 匵ã£ãŠããã¡ã ã£ãããã¥ãŒããªã¢ã«ãèŠãã倧åã¯1æéã§è§£æ±ºã§ãããã¥ãŒããªã¢ã«ãèŠãããšã«ãªããŸããããããã§è§£ããªã課é¡ã¯ãããŸããã§ãããæéã¯ãããã«ããŠããè³ã¿ããæ±ãæ»ãã»ã©æ©ã¿æããæéã¯çµ¶å¯Ÿã«èªåãè£åããŸãããåã¯ä»ãŸã§36幎ã³ã³ãã¥ãŒã¿ãŒæ¥çã§åããŠããŠããã£ãšãã®ããæ¹ã§åã«é²ãã§ããŠè£åãããããšã¯äžåºŠããããŸããã®ã§ãæ¥ãã°åãã®ç²Ÿç¥ã§ä¹ãåããŸãããããšã棺桶ã«ããã¶ããããã¯å ¥ã£ãŠãŠãããããªãæ³ã ãšããŠã^^;
çµå±ã©ãã ã£ãã®ïŒ
ãããããŸãããããããããã£ããšãããšãããäžçªã®ãã€ã³ãã§ãããæç³»ãã¢ã©éã§æ©æ¢°åŠç¿ããŸã ãŸã æ°ãããã®ã远ããããŠããããªãšããããšãåãããŸããããããŠãAIãšã人工ç¥èœãšãã·ã³ã®ã¥ã©ãªãã£ãšãã®èšèã«èžããããªããªã£ãããªãæ©æ¢°åŠç¿ã®æ°åŒãšã³ãŒããèŠãŠãéããïŒå€§ããªå£°ã§ã¯èšããŸãããïŒ**ãã·ã³ã®ã¥ã©ãªãã£ïŒããããïŒã**ãšæãããã«ãªããŸãããç ç©¶è ã¬ãã«ã®ãšããã§äœãè¡ãããŠããã®ãç¥ããŸããããå®è£ ã¬ãã«ã§ã¯æ©æ¢°åŠç¿ã«å¯ŸããŠãè±çã§è¶ã ãšæ¯ãããããªå¹»æ³ãæ±ããªããªããŸããã
äžããããããŒã¿ã»ãããšæ°åŒã«åŸã£ãŠãå»¶ã ãšèšç®ããŠããã ãã®ã³ã³ãã¥ãŒã¿ã·ã¹ãã ã§ããããæ©æ¢°åŠç¿ã¯ããã ã人éã«ã¯ãªãæ ¹æ°ãšéäžåã§äžãããã課é¡ãããã°ã©ã ãšããŠé£œããã«ç¹°ãè¿ããç¹ã ãã¯äººéãé¥ãã«åé§ããŠããŸãããã®èŸºãã¯æãªããã®ã³ã³ãã¥ãŒã¿ã·ã¹ãã ãšå€ãããããŸããããããã人éã«ãšã£ãŠæªç¥ã®ç¥èŠãçºèŠããããã®ããŒã«ãšããŠåœ¹ç«ã£ãŠãç¹ããåŸæ¥ã®æ¥åã·ã¹ãã ãšéããã®ãªã®ããªãšããã®ãã1ã¶æéæ©æ¢°åŠç¿ãå匷ããŠããåã®ææ³ã§ãã
æ°ããªæµ·åãèŠããŠããã®ã§ãæ¥å¹ŽåºŠããã®æ°ãããããžã§ã¯ãé 匵ããŸããæ©æ¢°åŠç¿ã®å匷ãçãããæ®ºããèªå次第ãå®å¹ŽãŸã§ã®1幎åãèª å®ã«æ¥œããæããååãã«åãçµãã§ãããŸãã
æåŸã«ãæ¹å€ãæããèšãããŠãããã°ãæ¬ å46ã¯å¥³æ§ã¢ã€ãã«çã«èãéããçŸä»£ã®å°ŸåŽè±ã ãšããæãããã®ã²ãšæã§ããã«åŒ·ããªããŸãããå»å¹Žã®çŽ çœã§ã®å§¿ã«ã¯ãæ¥æ¯è°·éé³ã§PAããé£ã³éããŠéªšæãã圌ã®å§¿ãéãªããŸãã
åããã¯ä»¥äžã§ããã€ããããã€ãã
