人工知能と機械学習 / AI and ML

Slide 1

Slide 1 text

ஜ೾େֶ৘ใՊֶྨഅ৔ઇ೫ CBCB!DTUTVLVCBBDKQ ஜ೾େֶڞ௨Պ໨ʢ৘ใʣ  ʮσʔλαΠΤϯεʯ  ਓ޻஌ೳͱػցֶश

Slide 2

Slide 2 text

/39 ຊߨٛͰֶͿ͜ͱ w ਓ޻஌ೳ͕Ͳ͏͍͏৔໘Ͱ׆༻͞Ε͍ͯΔͷ͔ w ͍·ͷਓ޻஌ೳʹ͸Կ͕Ͱ͖Δͷ͔ w ػցֶशͰ͸ͲͷΑ͏ʹσʔλ͔ΒͷֶशΛߦ͏ͷ͔ ਓ޻஌ೳʹԿ͕Ͱ͖Δͷ͔ɾػցֶश͸Ͳ͏͍͏͘͠Έ͔ 2

Slide 3

Slide 3 text

ਓ޻஌ೳ

Slide 4

Slide 4 text

/39 ਓ޻஌ೳͱ͸ w ਓ޻஌ೳ͸  ʮ஌తͳػցɺಛʹɺ஌తͳίϯϐϡʔλϓϩάϥϜΛ࡞ΔՊֶͱٕज़ʯ w ैདྷͷίϯϐϡʔλϓϩάϥϜ͸஌తͰ͸ͳ͔ͬͨ w ਓ͕ؒ஌ೳΛ࢖ͬͯͰ͖Δͷͱಉ͜͡ͱΛ  ίϯϐϡʔλϓϩάϥϜͰ΋Ͱ͖ΔΑ͏ʹ͍ͨ͠ ਓ޻஌ೳ͸ʮ஌తͳίϯϐϡʔλϓϩάϥϜʯ 4 IUUQTXXXBJHBLLBJPSKQXIBUTBJ"*GBRIUNM (*)

Slide 5

Slide 5 text

/39 ਓ޻஌ೳͱ͸ w ஌తͰ͸ͳ͍ίϯϐϡʔλϓϩάϥϜ͸  ༩͑ΒΕͨϧʔϧͲ͓Γͷಈ࡞͔͠Ͱ͖ͳ͍ ⿞ίϯϐϡʔλʹѻ͑ΔܗࣜͰਓ͕ؒϧʔϧΛ࡞Δඞཁ͕͋Δ ⿞ෳࡶͳ໰୊Ͱ͸͢΂ͯͷঢ়گΛ໢ཏͨ͠ϧʔϧΛ࡞Δ͜ͱ͸ࠔ೉ w ஌తͳίϯϐϡʔλϓϩάϥϜʢਓ޻஌ೳʣ͸  ঢ়گʹԠͯ͡ͲͷΑ͏ʹಈ࡞͢Ε͹ྑ͍͔ΛྟػԠมʹ൑அ͢Δ ⿞ܦݧ΍ڭࡐ͔ΒϧʔϧΛֶश͢Δ ⿞গ਺ͷϧʔϧ͔Βਪ࿦ͯ͠൑அ͢Δ ਓ޻஌ೳ͸ঢ়گʹԠͯ͡ྟػԠมʹಈ࡞͢Δ 5

Slide 6

Slide 6 text

/39 w ஌తͰ͸ͳ͍ίϯϐϡʔλϓϩάϥϜ͸ϧʔϧʹैͬͯೝࣝ͢Δ w ༷ʑͳච੻ͷखॻ͖จࣈΛ໢ཏͨ͠ϧʔϧΛ࡞Δͷ͸ࠔ೉    ਓ޻஌ೳͱ͸ ྫɿखॻ͖਺ࣈจࣈΛೝࣝ͢ΔίϯϐϡʔλϓϩάϥϜ 6 खॻ͖਺ࣈจࣈͷग़యɿIUUQZBOOMFDVODPNFYECNOJTU 1 1 ʮࠨӈࡾ෼ͷҰͷྖҬ͕ന৭ͳΒʯ

Slide 7

Slide 7 text

/39 ਓ޻஌ೳ͸ڭࡐʢը૾ͱ਺ࣈͷϖΞʣΛ࢖ͬͯೝࣝϧʔϧΛֶश͢Δ ਓ޻஌ೳͱ͸ ྫɿखॻ͖਺ࣈจࣈΛೝࣝ͢ΔίϯϐϡʔλϓϩάϥϜ 7 1 1 9 9 7 7 ϧʔϧΛֶश 9 ϧʔϧΛར༻ ڭࡐ ʜ खॻ͖਺ࣈจࣈͷग़యɿIUUQZBOOMFDVODPNFYECNOJTU

Slide 8

Slide 8 text

/39 ೔ৗͷதͷਓ޻஌ೳ w εϚʔτεϐʔΧʔɿ  ਓؒͷԻ੠ࢦࣔʹै͍༷ʑͳಈ࡞Λߦ͏ w ϩϘοτ૟আػɿ  ো֐෺Λආ͚ͳ͕Β෦԰શମΛ૟আ͢Δ w إೝূɿ  Χϝϥલͷਓ෺͕ొ࿥͞ΕͨਓͱಉҰਓ෺͔Ͳ͏͔Λ൑ఆ w ίϯςϯπɾ঎඼ਪનɿ  Ӿཡཤྺ΍ߪങཤྺʹ΋ͱ͖ͮϢʔβ͕޷ΉΞΠςϜΛఏࣔ զʑͷ೔ৗͷ͞·͟·ͳ৔໘Ͱਓ޻஌ೳ͕׆༻͞Ε͍ͯΔ 8

Slide 9

Slide 9 text

/39 ਓؒΛ௒͑Δਓ޻஌ೳ w *#.8BUTPOɿ  ΞϝϦΧͷΫΠζ൪૊ʮ+FPQBSEZʯʹ௅ઓɺৗ࿈ग़ԋऀΛഁΓ༏উ w ౦ϩϘ͘Μɿ  େֶೖࢼ໛ࢼͷ਺ֶɾੈք࢙Ͱภࠩ஋௒͑Λୡ੒ w %FFQ#MVFɿ  νΣεͷੈքνϟϯϐΦϯʹউར w "MQIB(Pɿ  ғޟͷੈքτοϓع࢜ʹউར ΫΠζ΍ήʔϜͰਓؒΛ௒͑Δਓ޻஌ೳ͕ొ৔ 9

Slide 10

Slide 10 text

/39 ݱࡏͷਓ޻஌ೳʹͰ͖Δ͜ͱɿಛԽܕWT൚༻ܕ w ਓ޻஌ೳ͸ʮಛԽܕʯͱʮ൚༻ܕʯͷೋछྨʹେผ͞ΕΔ w ಛԽܕਓ޻஌ೳɿ  ͋Δಛఆͷঢ়گ΍໰୊ʹ͓͍ͯ஌తʹ;Δ·͏ਓ޻஌ೳ w ൚༻ܕਓ޻஌ೳɿ  ਓؒͱಉ༷ʹ༷ʑͳঢ়گɾ໰୊ʹ͓͍ͯ஌తͳ;Δ·͍͕Ͱ͖Δਓ޻஌ೳ w ݱࡏͷਓ޻஌ೳ͸ಛԽܕਓ޻஌ೳͰ͋Γ  ൚༻ܕਓ޻஌ೳ͸ະ࣮ͩݱ͞Ε͍ͯͳ͍  ݱࡏͷਓ޻஌ೳ͸ಛఆͷ͜ͱ͚ͩͰ͖ΔಛԽܕਓ޻஌ೳ 10

Slide 11

Slide 11 text

/39 ݱࡏͷਓ޻஌ೳʹͰ͖Δ͜ͱɿը૾ॲཧ w ը૾ೝࣝɿ  ࣸਅʹ͍ࣸͬͯΔ΋ͷΛೝࣝ͢Δ w ෺ମݕग़ɿ  ࣸਅͷͲ͜ʹԿ͕͍ࣸͬͯΔͷ͔Λೝࣝ͢Δ w ը૾ੜ੒ɿ  ຊ෺ͷࣸਅͷΑ͏ͳը૾Λਓ޻తʹ࡞Γग़͢ ಛԽܕਓ޻஌ೳͰ΋༷ʑͳ͜ͱ͕࣮ݱͰ͖Δ 11 ग़యɿIUUQDTOTUBOGPSEFEVTMJEFTDTO@@MFDUVSFQEG  ग़యɿIUUQTBSYJWPSHBCT ຊ෺ͷࣸਅ ਓ޻తͳࣸਅ (*) (*)

Slide 12

Slide 12 text

/39 ݱࡏͷਓ޻஌ೳʹͰ͖Δ͜ͱɿݴޠॲཧ w จॻ෼ྨɿ  จॻΛಛఆͷΧςΰϦʹ෼ྨ͢Δ w ػց຋༁ɿ  ͋ΔݴޠͰॻ͔ΕͨจষΛผͷݴޠʹ຋༁͢Δ w ৘ใݕࡧɿ  ಛఆͷ৘ใ͕ܝࡌ͞Ε͍ͯΔ΢ΣϒϖʔδΛݟ͚ͭΔ w ࣭໰Ԡ౴ɿ  จষͰ༩͑ΒΕ࣭ͨ໰ʹର͢Δ౴͑Λฦ͢ ಛԽܕਓ޻஌ೳͰ΋༷ʑͳ͜ͱ͕࣮ݱͰ͖Δ 12

Slide 13

Slide 13 text

/39 ݱࡏͷਓ޻஌ೳʹͰ͖Δ͜ͱɿԻ੠ॲཧ w Ի੠ೝࣝɿ  ࿩͞Ε͍ͯΔ಺༰ΛจষͰॻ͖ى͜͢ w ࿩ऀಉఆɿ  ొ࿥͞Εͨਓ෺ͱಉ͡ਓ͕࿩͍ͯ͠Δ͔൑ఆ͢Δ w Ի੠߹੒ɿ  ਓ͕ؒ࿩͍ͯ͠ΔΑ͏ͳ੠Λਓ޻తʹ࡞Γग़͢ ಛԽܕਓ޻஌ೳͰ΋༷ʑͳ͜ͱ͕࣮ݱͰ͖Δ 13

Slide 14

Slide 14 text

/39 ݱࡏͷਓ޻஌ೳʹͰ͖Δ͜ͱɿϩϘοτ w ؀ڥೝࣝɿ  ηϯα৘ใ౳͔ΒपғʹԿ͕͋Δ͔Λೝࣝ͢Δ w ܦ࿏ɾߦಈܭըɿ  Ҡಈɾಈ࡞ࢦࣔʹै͏ͨΊʹͲͷΑ͏ͳܦ࿏Λ௨Δ͔ɺ  Ͳͷ෦෼ΛͲͷΑ͏ʹಈ͔͔͢Λܭը͢Δ ಛԽܕਓ޻஌ೳͰ΋༷ʑͳ͜ͱ͕࣮ݱͰ͖Δ 14

Slide 15

Slide 15 text

/39 ਓ޻஌ೳͷ׆༻ྫ ྫᶃεϚʔτεϐʔΧʔ ಛԽܕਓ޻஌ೳͷ૊߹ͤͰ͞Βʹߴ౓ͳ͜ͱΛ࣮ݱͰ͖Δ 15 4UFQىಈΩʔϫʔυͷೝࣝ 4UFQԻ੠ʹΑΔࢦࣔΛจষͱͯ͠ೝࣝ 4UFQࢦࣔจ͔Β໨తΛೝࣝ 4UFQࢦࣔΛ࣮ߦ

Slide 16

Slide 16 text

/39 ਓ޻஌ೳͷ׆༻ྫ ྫᶄࣗಈӡస ಛԽܕਓ޻஌ೳͷ૊߹ͤͰ͞Βʹߴ౓ͳ͜ͱΛ࣮ݱͰ͖Δ 16 ը૾ͷग़యɿIUUQTTUPSBHFHPPHMFBQJTDPNTEDQSPEWTBGFUZSFQPSUXBZNPTBGFUZSFQPSUQEG 4UFQଞͷं྆΍าߦऀ౳ͷݕ஌ 4UFQଞͷं྆΍าߦऀ౳ͷߦಈ༧ଌ 4UFQ໨త஍·Ͱͷܦ࿏Λܭը 4UFQܦ࿏ʹै͏ͨΊͷಈ࡞Λܭը

Slide 17

Slide 17 text

/39 ਓ޻஌ೳͷ·ͱΊ w ਓ޻஌ೳ͸஌తͳίϯϐϡʔλϓϩάϥϜͰ  ϧʔϧΛֶशɾਪ࿦ͯ͠ྟػԠมʹಈ࡞Ͱ͖Δ w ͍·ͷਓ޻஌ೳ͸ಛԽܕͰ͋Δಛఆͷ໰୊͔͠ղ͚ͳ͍ w ༷ʑͳ෼໺ͰಛԽܕਓ޻஌ೳͷݚڀ͕ਐΊΒΕ  ͦΕΒͷ૊߹ͤͰΑΓߴ౓ͳਓ޻஌ೳ͕࣮ݱ͞Εͭͭ͋Δ ͍·ͷਓ޻஌ೳ͸ಛԽܕ͕༷ͩʑͳ͜ͱ͕Ͱ͖Δ 17

Slide 18

Slide 18 text

ػցֶश

Slide 19

Slide 19 text

/39 ػցֶशͱ͸ w ػցֶश͸  ίϯϐϡʔλϓϩάϥϜʹσʔλ͔ΒϧʔϧΛֶशͤ͞Δٕज़ w େྔͷೖग़ྗྫͷσʔλΛڭࡐͱͯ͠༩͑  ೖྗͱग़ྗΛରԠ͚ͮΔϧʔϧΛֶशͤ͞Δ ೖྗͱग़ྗΛରԠ͚ͮΔϧʔϧΛֶश͢Δ 19 ຊߨٛͰऔΓ্͛Δػցֶश͸ɺݫີʹ͸ʮڭࢣ෇͖ػցֶशʯͱݺ͹ΕΔ 7 ೖྗ ग़ྗ ը૾ ਺ࣈ

Slide 20

Slide 20 text

/39 ػցֶशͱ͸ ೖྗͱग़ྗΛରԠ͚ͮΔϧʔϧΛֶश͢Δ 20 The quick brown fox jumps over the lazy dog ૉૣ͍஡৭ͷޅ͸ ͷΖ·ͳݘΛඈͼ ӽ͑Δ “Hello, world!” ೖྗ ग़ྗ ݈߁ϦεΫɿߴ ग़యɿIUUQDTOTUBOGPSEFEVTMJEFTDTO@@MFDUVSFQEG  ग़యɿIUUQTNBHFOUBUFOTPSqPXPSHOTZOUIGBTUHFO จষ จষ ը૾ ෺ମͱҐஔ Ի੠ จষ ױऀ৘ใ ݈߁ϦεΫ (*) (**)

Slide 21

Slide 21 text

/39 ػցֶशͱ͸ w εϚʔτϑΥϯɾηϯαɾΠϯλʔωοτɾ΢ΣϒαʔϏε౳ͷීٴͰ  σʔλ͕૿େͨ͠ͷ͕ػցֶशͷൃలͷ͖͔͚ͬ w ιʔγϟϧϝσΟΞ΁ͷ౤ߘ΍  ΢ΣϒαʔϏε্Ͱͷਓʑͷߦಈϩά΋ػցֶशͷڭࡐͱͯ͠༻͍ΒΕΔ w ޿ࠂ഑৴ͳͲͰ΢ΣϒαʔϏεӡӦاۀ͸རӹΛ্͛ɺ  ແྉͰ΢ΣϒαʔϏεΛఏڙ͠ͳ͕Βڭࡐ༻ͷσʔλΛऩू͍ͯ͠Δ ͋ΒΏΔ΋ͷ͕σʔλԽ͞ΕػցֶशͰ࢖ΘΕΔ 21

Slide 22

Slide 22 text

/39 ػցֶशͷ͘͠Έɿ܇࿅σʔλ w ྫ୊ɿʮजᙾ͕ྑੑ͔ѱੑ͔ʯ൑ఆ͢ΔͨΊͷϧʔϧΛֶशͤ͞Δ w ڭࡐͱͯ͠༻͍ΔσʔλΛ܇࿅σʔλͱݺͿ w ֶशͨ͠ϧʔϧΛ༻͍ͯ  ܇࿅σʔλʹͳ͍जᙾ͕ྑੑ͔ѱੑ͔Λਖ਼͘͠൑ఆͰ͖ΔΑ͏ʹ͍ͨ͠  ೖग़ྗྫΛ܇࿅σʔλͱͯ͠༻͍ͯϧʔϧΛֶश͢Δ 22 जᙾը૾ͷग़యɿW. N. Street et al. , Nuclear feature extraction for breast tumor diagnosis. Proc. of the International Symposium on Electronic Imaging, 1993. ྑੑ ѱੑ ྑੑ ѱੑ ܇࿅σʔλ ೖྗʢजᙾʣΛ  ग़ྗʢྑੑPSѱੑʣʹ  ରԠ͚ͮΔ  ϧʔϧ ֶश

Slide 23

Slide 23 text

/39 ػցֶशͷ͘͠Έɿ܇࿅σʔλ w ίϯϐϡʔλͰܭࢉͰ͖ΔΑ͏ʹ܇࿅σʔλ͸਺஋Ͱදݱ͞Ε͍ͯΔ w ྫɿʮେ͖͞ʯͱʮͰ͜΅͜౓ʯͰजᙾΛදݱ͢Δ  ʢԿΒ͔ͷखஈͰ਺஋Խ͞Ε͍ͯΔͱ͢Δʣ ܇࿅σʔλ͸਺஋Ͱදݱ͞Εͳ͍ͱ͍͚ͳ͍ 23 ܇࿅σʔλͷྫ ೖྗ ग़ྗ େ͖͞ Ͱ͜΅͜౓ जᙾ 0.252 0.014 ྑੑ जᙾ 0.327 0.340 ྑੑ जᙾ 1.000 0.371 ѱੑ जᙾ 0.223 0.369 ѱੑ ʜ ʜ ʜ ʜ

Slide 24

Slide 24 text

/39 ػցֶशͷ͘͠Έɿ܇࿅σʔλ ܇࿅σʔλ͸਺஋Ͱදݱ͞Εͳ͍ͱ͍͚ͳ͍ 24 211 220 223 … 213 217 217 220 … 210 214 217 216 … 205 … … … … … 216 217 214 … 181 The quick brown fox jumps over the lazy dog aardvark … brown … dog … zzzat 0 … 1 … 1 … 0

Slide 25

Slide 25 text

/39 ػցֶशͷ͘͠Έɿ܇࿅σʔλ ܇࿅σʔλ͸਺஋Ͱදݱ͞Εͳ͍ͱ͍͚ͳ͍ 25 ܇࿅σʔλͷྫ ೖྗ ग़ྗ େ͖͞ Ͱ͜΅͜౓ जᙾ 0.252 0.014 ྑੑ जᙾ 0.327 0.340 ྑੑ जᙾ 1.000 0.371 ѱੑ जᙾ 0.223 0.369 ѱੑ ʜ ʜ ʜ ʜ ਤࣔ େ͖͞ Ͱ͜΅͜౓ ྑੑͷजᙾ ѱੑͷजᙾ जᙾ जᙾ

Slide 26

Slide 26 text

/39 ػցֶशͷ͘͠Έɿϧʔϧ w ίϯϐϡʔλ͕ܭࢉͰ͖ΔΑ͏ʹϧʔϧ΋਺ࣜͰදݱ͞ΕΔ w ୯७ͳϧʔϧͷܗࣜɿ ϧʔϧ΋਺ࣜͰදݱ͞Εͳ͍ͱ͍͚ͳ͍ 26 େ͖͞ Ͱ͜΅͜౓   ͳΒྑੑɺͦΕҎ֎ͳΒѱੑ z = w1 × + w2 × + b; z < 0

Slide 27

Slide 27 text

/39 ػցֶशͷ͘͠Έɿϧʔϧ ϧʔϧ΋਺ࣜͰදݱ͞Εͳ͍ͱ͍͚ͳ͍ 27 େ͖͞ Ͱ͜΅͜౓ ྑੑͷजᙾ ѱੑͷजᙾ [ʾͷྖҬ [ͷྖҬ ͷ৔߹ɿ  େ͖͞ Ͱ͜΅͜౓ w1 = 1, w2 = − 1, b = 0 z = − େ͖͞ Ͱ͜΅͜౓ z = w1 × + w2 × + b;

Slide 28

Slide 28 text

/39 ػցֶशͷ͘͠Έɿϧʔϧ X X Cͷ஋͕มΘΔͱϧʔϧ΋มΘΔ 28 େ͖͞ Ͱ͜΅͜౓ ྑੑͷजᙾ ѱੑͷजᙾ [ʾͷྖҬ [ͷྖҬ ͷ৔߹ɿ  େ͖͞ Ͱ͜΅͜౓ w1 = 1, w2 = 1, b = − 1 z = + −1 େ͖͞ Ͱ͜΅͜౓ z = w1 × + w2 × + b;

Slide 29

Slide 29 text

/39 ػցֶशͷ͘͠Έɿϧʔϧ ܇࿅σʔλʹ౰ͯ͸·Δྑ͍ϧʔϧΛֶश͍ͨ͠ 29 େ͖͞ Ͱ͜΅͜౓ େ͖͞ Ͱ͜΅͜౓ w1 = 1, w2 = − 1, b = 0 w1 = 1, w2 = 1, b = − 1 ϧʔϧ͕౰ͯ͸·Βͳ͍जᙾ ӈͷϧʔϧͷํ͕ɺ܇࿅σʔλʹ౰ͯ͸·Δ

Slide 30

Slide 30 text

/39 ػցֶशͷ͘͠Έɿଛࣦ ϧʔϧͷྑ͠ѱ͠Λ఺਺Ͱද͢ 30 ଛࣦɿେ ଛࣦɿখ େ͖͞ Ͱ͜΅͜౓ େ͖͞ Ͱ͜΅͜౓ w ଛࣦɿϧʔϧͷѱ͞Λ఺਺Ͱදͨ͠΋ͷ w ܇࿅σʔλʹ౰ͯ͸·Βͳ͍ϧʔϧ͸ଛࣦ͕େ͖͘ͳΔ

Slide 31

Slide 31 text

/39 ػցֶशͷ͘͠Έɿଛࣦ w ʮޡ൑ఆͷ݅਺ʯΛଛࣦͱͯ͠࢖͏ͱϧʔϧͷࡉ͔͍ҧ͍͕Θ͔Βͳ͍ w ϧʔϧʹʮѱੑͷ֬཰ʯΛग़ྗͤ͞ɺͦΕʹ΋ͱ͖ͮଛࣦΛࢉग़ ϧʔϧͷྑ͠ѱ͠Λ఺਺Ͱද͢ 31 ૯࿨Λϧʔϧͷଛࣦͱ͢Δ ଛࣦɿ ܇࿅σʔλ [ ѱੑͷ֬཰  ଛࣦ ೖྗ ग़ྗ େ͖͞ Ͱ͜΅͜౓ जᙾ 0.252 0.014 ྑੑ -0.542 0.336 0.410 जᙾ 0.327 0.340 ྑੑ -0.172 0.457 0.611 जᙾ 1.000 0.371 ѱੑ 1.208 0.770 0.261 जᙾ 0.223 0.369 ѱੑ -0.348 0.414 0.882 ʜ ʜ ʜ ʜ a = σ(z)

Slide 32

Slide 32 text

/39 ػցֶशͷ͘͠Έɿޯ഑๏ w ଛࣦ͕࠷খͷϧʔϧɺ  ͭ·Γ܇࿅σʔλʹ࠷΋౰ͯ͸·ΔϧʔϧΛݟ͚ͭΔͷֶ͕श w ޮ཰తʹͦͷΑ͏ͳϧʔϧΛݟ͚ͭΔͨΊʹޯ഑๏͕༻͍ΒΕΔ w ޯ഑๏Ͱ͸ɺଛࣦ͕࠷΋খ͘͞ͳΔํ޲ʹগ͚ͩ͠ਐΉ  ʢX X Cͷ஋Λগ͚ͩ͠มԽͤ͞Δʣ͜ͱΛ܁Γฦ͢ ଛࣦ͕࠷খͱͳΔϧʔϧʢX X Cͷ஋ʣΛݟ͚ͭΔ 32

Slide 33

Slide 33 text

/39 ػցֶशͷ͘͠Έɿ൚Խೳྗ w ֶशͷͦ΋ͦ΋ͷ໨త͸  ʮ܇࿅σʔλʹͳ͍ະ஌ͷೖྗʹରͯ͠ਖ਼͍͠ग़ྗΛฦ͢ʯϧʔϧͷֶश w ܇࿅σʔλ΁ͷ౰ͯ͸·Γ͚ͩΛߟྀ͢Δͱ  ϧʔϧ͕܇࿅σʔλʹ౰ͯ͸·Γ͗ͯ͢ະ஌ͷೖྗͰ͸ؒҧ͑ΔڪΕ w X Xͷ஋͕ۃ୺ʹେ͖͍ϧʔϧ͸ա৒ద߹͍ͯ͠ΔڪΕ͕͋ΔͷͰ  ͦͷΑ͏ͳϧʔϧʹϖφϧςΟΛ༩͑ΔΑ͏ʹଛࣦΛઃܭ͢Δ w ʮϧʔϧ͕ະ஌ͷೖྗʹର͠ਖ਼͍͠ग़ྗΛฦ͢ೳྗʯΛ  ൚Խೳྗͱ͍͏ ܇࿅σʔλʹ౰ͯ͸·Γ͗͢Δϧʔϧ͸൚Խೳྗ͕௿͍ 33

Slide 34

Slide 34 text

/39 w ௚ઢͰදݱ͞ΕΔϧʔϧ͸ೖྗͱग़ྗͷରԠ͚͕ͮෳࡶͳ৔߹ʹෆे෼ w ਂ૚ֶशΛ༻͍ΔͱෳࡶͳϧʔϧΛֶशͰ͖Δ େ͖͞ Ͱ͜΅͜౓ େ͖͞ Ͱ͜΅͜౓ ୯७ͳϧʔϧ ػցֶशͷ͘͠Έɿਂ૚ֶश ෳࡶͳϧʔϧͷֶशʹ༻͍ΒΕΔͷ͕ਂ૚ֶश 34 େ͖͞ Ͱ͜΅͜౓ ਂ૚ֶशʹΑΔϧʔϧ

Slide 35

Slide 35 text

/39 w ୯७ͳϧʔϧͷ໛ࣜਤɿ ػցֶशͷ͘͠Έɿਂ૚ֶश ਂ૚ֶशͰ͸ෳ਺ͷඇઢܗؔ਺Λଟ૚ʹॏͶΔ 35 େ͖͞ Ͱ͜΅͜౓ ྑੑ·ͨ͸ѱੑ w1 w2 a=σ(z) େ͖͞ Ͱ͜΅͜౓ ྑੑ·ͨ͸ѱੑ w ਂ૚ֶशʹΑΔϧʔϧͷ໛ࣜਤɿ

Slide 36

Slide 36 text

/39 ػցֶशͷ࣮૷ w ػցֶशͷ࣮૷ʹ͸ɺϓϩάϥϛϯάݴޠ1ZUIPO͕޿͘࢖ΘΕ͍ͯΔ w ྫ͑͹TDJLJUMFBSOͱ͍͏1ZUIPOϥΠϒϥϦΛ࢖͏ͱɺ  ਺ߦͷίʔυͰػցֶशΛ࣮૷͢Δ͜ͱ͕Ͱ͖Δ w ͍͔ͭ͘ͷαϯϓϧσʔλ΋TDJLJUMFBSOͰఏڙ͞Ε͍ͯΔ ༷ʑͳϥΠϒϥϦ͕ެ։͞Ε͓ͯΓ؆୯ʹ࣮૷͕Ͱ͖Δ 36 https://scikit-learn.org/ ϧʔϧͷֶश ϧʔϧͷద༻

Slide 37

Slide 37 text

/39 ػցֶशͷ࣮૷ w ਂ૚ֶशͷ1ZUIPOϥΠϒϥϦ΋༷ʑͳ΋ͷ͕ఏڙ͞Ε͍ͯΔɿ  5FOTPS'MPX ,FSBT 1Z5PSDI $IBJOFS w (PPHMF$PMBCPSBUPSZΛ࢖͏ͱԾ૝ϚγϯΛ࢖ͬͯ  ਂ૚ֶशϥΠϒϥϦΛ΢Σϒϒϥ΢β͔Β؆୯ʹ࢖͏͜ͱ͕Ͱ͖Δ ༷ʑͳϥΠϒϥϦ͕ެ։͞Ε͓ͯΓ؆୯ʹ࣮૷͕Ͱ͖Δ 37 https://colab.research.google.com

Slide 38

Slide 38 text

/39 ػցֶशͷ࣮૷ w ,BHHMFͳͲͷػցֶशίϯϖςΟγϣϯͰ͸  ࣮ફతͳػցֶशͷ՝୊ɾσʔλ͕ఏڙ͞Εଞऀͱڝ͏͜ͱ͕Ͱ͖Δ ػցֶशίϯϖςΟγϣϯͰྗࢼ͠Λ͢Δ͜ͱ͕Ͱ͖Δ 38 https://www.kaggle.com/c/titanic/leaderboard

Slide 39

Slide 39 text

/39 ػցֶशͷ·ͱΊ w େྔͷೖग़ྗྫͷσʔλΛ܇࿅σʔλͱͯ͠࢖͍  ೖྗͱग़ྗΛରԠ͚ͮΔϧʔϧΛֶश w ܇࿅σʔλʹͰ͖Δ͚ͩ౰ͯ͸·ΔΑ͏ͳϧʔϧ  ʢଛࣦͷখ͍͞ϧʔϧʣΛޯ഑๏Λ༻͍ͯݟ͚ͭΔ ܇࿅σʔλΛ༻͍ͯೖྗͱग़ྗΛରԠ͚ͮΔϧʔϧΛֶश 39