AIを一つください、という前に

Slide 1

Slide 1 text

AIΛͻͱ͍ͭͩ͘͞ɺͱ͍͏લʹ ͕͜͜มͩΑɺ೔ຊͷAIϓϩδΣΫτ

Slide 2

Slide 2 text

໨࣍ 1.ࣗݾ঺հ 2.ຊ೔ͷओࢫ 3.ػցֶशϓϩδΣΫτͰ௚໘ͨ͠՝୊ 4.·ͱΊ

Slide 3

Slide 3 text

ࣗݾ঺հ • ਿࢁ Ѩ੟ • iOSΤϯδχΞग़਎ • ࣾ಺ϔϧϓγεςϜ༻ʹBotΛ࡞੒ • ݱ AIϦαʔνϟʔ (2019/01ʙ)

Slide 4

Slide 4 text

ݟͯࢼͯ͠Θ͔ΔػցֶशΞ ϧΰϦζϜͷ͘͠Έ ػցֶशਤؑ • ஶऀͷҰਓͱͯ͠ࢀՃ(3ਓͷڞஶ) • ෳࡶͳػցֶशΞϧΰϦζϜͷ࢓૊Έ ΛҰ࡭Ͱֶ΂Δ • ػցֶशͷਤΛͨ͘͞Μܝࡌ • ίʔυΛهࡌ͍ͯ͠ΔͷͰɺݟΔ͚ͩ Ͱͳ͘ࢼ͢͜ͱ΋Ͱ͖Δ

Slide 5

Slide 5 text

ΞϯέʔτΛ͍ͤͯͩ͘͞͞

Slide 6

Slide 6 text

AIʹ͍ͭͯฉ͍ͨ͜ͱ͕͋Δ

Slide 7

Slide 7 text

AIʹ͍ͭͯݕ౼͍ͯ͠Δ

Slide 8

Slide 8 text

AIΛݱ৔ʹಋೖ͍ͯ͠Δ

Slide 9

Slide 9 text

Զ͕AIͩ

Slide 10

Slide 10 text

AIͷഎޙʹ͸ਓ͕͍Δ • େྔͷσʔλʹ͍ͭͯख࡞ۀͰ෼ྨΛߦ͏Ξϊςʔλʔ • KPIΛఆٛͯ͠؂ࢹ͢ΔϏδωεͷεϖγϟϦετ • σʔλΛಧ͚ΔύΠϓϥΠϯͷߏஙͱอकΛߦ͏ΤϯδχΞ • ػցֶशϞσϧͷ࡞੒ͱϝϯςφϯεΛߦ͏αΠΤϯςΟετ

Slide 11

Slide 11 text

ͦͯ͠஍ຯʹటष͍ • ख࡞ۀͰ෼ྨͨ݁͠ՌΛμϒϧνΣοΫ • ໨ඪઃఆʹ͍ͭͯݱ৔ͱലେͳίϛϡχέʔγϣϯ • σʔλ͕૝ఆ֎ͷܗࣜͩͬͨͨΊʹൃੜ͢Δ੔ܗ࡞ۀ • จࣈ਺͕38จࣈΛӽ͑Δ৔߹ෳ਺Ϩίʔυʹ෼ׂɺͳͲ1 • Python΍SQLͰಛ௃ྔΤϯδχΞϦϯάͱ͍͏໊ͷඇਖ਼نԽ 1 ༣ศ൪߸σʔλͷઆ໌ - ೔ຊ༣ศ, Image by Joseph Emmanuel Japos from Pixabay

Slide 12

Slide 12 text

ຊ೔ͷओࢫ • ൒ܘ5mͰݟฉ͖ͨ͠͞·͟·ͳటष͍࿩ΛϑΣʔζ͝ͱʹڞ༗ ຊ೔࿩͞ͳ͍͜ͱ • టष͘ͳ͍͜ͱ • ࠷৽ͷݚڀ݁Ռ΍࠷৽ͷٕज़ಈ޲

Slide 13

Slide 13 text

ػցֶशϓϩδΣΫτ ͷϑΣʔζ෼͚ • ϑΣʔζ͸ʮ࢓ࣄͰ࢝ΊΔػցֶशʯ ͷ෼͚ํʹै͍·͢ 1.໰୊ఆٛ 2.Ծઆઃఆ 3.։ൃ 4.ޮՌݕূ

Slide 14

Slide 14 text

ػցֶशϓϩδΣΫτͰ௚໘ͨ͠՝୊ 1.໰୊ఆٛ <- 2.Ծઆઃఆ 3.։ൃ 4.ޮՌݕূ

Slide 15

Slide 15 text

໰୊ఆٛϑΣʔζ • ࣮ࢪࣄ߲ : AIͰऔΓ૊Έ͍ͨ໨తͷఆٛ • ۤ࿑࿩ 1-1. Ͳ͏࢝ΊΔͷ͔෼͔Βͳ͍ • ۤ࿑࿩ 1-2. ԿΛ΍Ε͹͍͍ͷ͔෼͔Βͳ͍

Slide 16

Slide 16 text

1-1. Ͳ͏࢝ΊΔͷ͔෼͔Βͳ͍ 1.AIͬͯԿ͔ษڧ͍ͨ͠ 2.AIͰԿ͕Ͱ͖Δ͔ݕূ͍ͨ͠ 3.AIͰ੒ՌΛ΋ͨΒ͍ͨ͠ • 3.Λ΍Γ͍͚ͨΕͲɺ1.͔Β͡Όͳ͍ͱ࢝ΊΒΕͳ͍ • ษڧ͠Α͏ͱ͢ΔͱՌͯ͠ͳ͍

Slide 17

Slide 17 text

How to Choose Your First AI Project3 1.୹ظؒͰখ͘͞੒ՌΛ͋͛ΒΕͦ͏͔ʁ 2.ن໛͕খ͗͢͞ͳ͍͔ʁେ͖͗͢ͳ͍͔ʁ 3.ࣗࣾͷઐ໳෼໺ͰͷऔΓ૊Έ͔ʁ 4.৴པͰ͖Δύʔτφʔͱ࢝ΊΒΕΔ͔ʁ 5.Ձ஋Λ૑Γग़ͤΔ͔ʁ 3 Andrew Ng: How to Choose Your First AI Project ༁͸චऀʹΑΔ

Slide 18

Slide 18 text

1-2. ԿΛ΍Ε͹͍͍ͷ͔෼͔Βͳ͍ • Internet, Smartphone͕௨ͬͨಓ • Կ͕Ͱ͖Δͷ͔ɺઐ໳ՈͰ͸ͳ͍ͷͰ෼͔Βͳ͍ • Կ͕Ͱ͖Δͷ͔ɺઐ໳ՈͰ΋͍·͍ͪ͸͖ͬΓ͠ͳ͍ • ϏδωεతʹԿ͕Ͱ͖Ε͹͍͍͔ɺ֎෦ͷઐ໳Ո͸෼͔Βͳ͍

Slide 19

Slide 19 text

AI Transformation Playbook2 1.੎͍Λ͚ͭΔͨΊύΠϩοτΛ࢝ΊΔ 2.ࣾ಺ʹAIνʔϜΛ࡞Δ 3.τϨʔχϯάΛࣾ಺ʹ޿͘ఏڙ͢Δ 4.AIઓུΛཱͯΔ 5.ࣾ಺ɾࣾ֎ͱͷίϛϡχέʔγϣϯύ εΛཱ֬͢Δ 2 AI Transformation Playbook How to lead your company into the AI era - Landing AI ༁͸චऀ

Slide 20

Slide 20 text

ػցֶशϓϩδΣΫτͰ௚໘ͨ͠՝୊ 1.໰୊ఆٛ 2.Ծઆઃఆ <- 3.։ൃ 4.ޮՌݕূ

Slide 21

Slide 21 text

2. ԾઆઃఆϑΣʔζ • ࣮ࢪࣄ߲ : AIͰར༻͢Δσʔλͱɺ༧ଌ͢Δର৅ͷఆٛ • ۤ࿑࿩2-1. ͦΕͬͯAIͳͷ • ۤ࿑࿩2-2. ʮσʔλ͸͋Δ(͋Δͱ͸ݴ͍ͬͯͳ͍)ʯ

Slide 22

Slide 22 text

2-1. ͦΕͬͯAIͳͷ • ϐϘοτςʔϒϧͰͳΜͱ͔ͳΔ৔߹͕͋Δ • ʮ͜ͷ͋ͨΓߦ͚ͦ͏ͳΜͰҰճ΍ͬͯΈ·͠ΐ͏ʯ →ʮͦΕ౷ܭͩΑͶʁʯ • ར༻Λ૝ఆ͍ͯ͠ΔΞϧΰϦζϜΛஸೡʹઆ໌͢Δͱಉ͜͡ͱ ΛݴΘΕΔ৔߹΋ (ϩδεςΟοΫճؼͱ͔) Image by Steve Buissinne from Pixabay

Slide 23

Slide 23 text

࣍ߦ͜͏࣍ʂ • AIΛ࢖ΘͣʹղܾͰ͖ΔͷͳΒͦΕͰOKͱͯ࣍͠ʹߦ͘ • ݱ৔ͷਓ͕ن४͸͋Δ͚ΕͲ೰Έͳ͕Β෼ྨ͍ͯ͠ΔɺͷΑ͏ ͳᐆດͳ՝୊ʹऔΓ૊Ή

Slide 24

Slide 24 text

2-2.ʮσʔλ͸͋Δ(͋Δͱ͸ݴ͍ͬͯͳ͍)ʯ Image by Angelo Giordano from Pixabay

Slide 25

Slide 25 text

σʔλͷอଘঢ়ଶ ൃੜ͕ͪ͠ͳ՝୊ ݱ৔ʹࢴͰอଘ͞Ε͍ͯΔ ར༻ෆՄ! Excelாථʹॻ͔Ε͍ͯΔ खಈͰͷ੔ܗ" جװγεςϜʹ͋Δ ಠࣗܗࣜσʔλ# ϩάͱͯ͠อ؅͞Ε͍ͯΔ ݱࡏͷσʔλͱҧ͏$ AIʹར༻Ͱ͖Δσʔλ͕͋Δ ݱ࣮ʹଘࡏ͠ͳ͍%

Slide 26

Slide 26 text

جװγεςϜͷಠࣗܗࣜσʔλ໰୊ • dateΧϥϜʹ"20190315"(จࣈྻ), "00000000"(จࣈྻ) • ઌਓ͕ۤ࿑Λ࣮ͯ͠૷ͯ͠ݱࡏಈ͍͍ͯΔ΋ͷ • ϦεϖΫτΛ๨Εͯ͸͍͚ͳ͍ ϩάͱͯ͠อ؅͞Ε͍ͯΔ΋ͷ͕ݱࡏͱҧ͏໰୊ • γεςϜӡ༻։࢝ޙʹݱ৔Ͱͷӡ༻ํ๏͕มΘΔͱൃੜ • ݸਓ৘ใอޢͷ؍఺͔Βݸਓ໊ΛϚεΫͯ͠ه࿥͢ΔɺͳͲ

Slide 27

Slide 27 text

ۀ຿γεςϜ͸ର৅ۀ຿ͷͨΊʹ͋Δ • ۀ຿γεςϜʹ͸͔֬ʹૉ੖Β͍͠σʔλ͕຾͍ͬͯΔ • ҰํɺAIͰ͸ࠓ·Ͱߟྀͯ͜͠ͳ͔ͬͨࣄ߲ʹऔΓ૊Ή • ಛʹϨΨγʔͳγεςϜͰ͸ʮAIͰ࢖͑ΔΑ͏ʹσʔλ͕อ؅ ͞ΕΔ͜ͱʯͱ͍͏ͷ͸ཁ݅ͱͯ͠ߟྀ͞Ε͍ͯͳ͍ • ͜Ε͔Β࡞ΔγεςϜͰ͸࠷ॳ͔ΒߟྀͰ͖Δͱϕλʔ

Slide 28

Slide 28 text

σʔλ͕͋Δͱ͍͏ཧ༝Ͱબ͹ͳ͍ • AIʹར༻Ͱ͖Δσʔλ͸ɺ࠷ॳ͸ͳ͍͜ͱ͕΄ͱΜͲ • ߴੑೳεϓϨουγʔτϚελʔ4͕ѻ͑Δن໛Ͱ࢝ΊΔ ...and it may be possible to build a valuable AI system with perhaps as few as 100-1000 data records (though more does not hurt).5 5 Andrew Ng: How to Choose Your First AI Project 4 ਺ઍߦͷεϓϨουγʔτΛݟͯ΋৺͕ંΕͳ͍ͱ͍͏ಛघٕೳΛ࣋ͬͨਓࡐ

Slide 29

Slide 29 text

ػցֶशϓϩδΣΫτͰ௚໘ͨ͠՝୊ 1.໰୊ఆٛ 2.Ծઆઃఆ 3.։ൃ <- 4.ޮՌݕূ

Slide 30

Slide 30 text

։ൃϑΣʔζ • PoCͰMVPΛ࡞੒͢Δ • ݁ՌΛαʔϏεԽ͢Δ • ۤ࿑࿩3-1. PoCͰͷࣾ಺ௐ੔ • ۤ࿑࿩3-2. αʔϏεԽ࣌ͷߟྀࣄ߲ଟ͗͢

Slide 31

Slide 31 text

3-1. PoCͰͷࣾ಺ௐ੔໰୊ • ॏཁͳ͜ͱͳͷͰ͖ͪΜͱ΍Δඞཁ͕͋Δ • ͜ͷσʔλΛ౉ͨ͢Ίͷ๏຿৹ࠪʹ1ϲ݄ɺܾࡋͷͨΊͷࣾ಺ௐ ੔ʹ਺ϲ݄͔͔Γ·͢ͱͳΔͱଈ࠲ʹ஗Ԇ͢Δ • લྫ͕ͳ͍ͷͰ൑அͰ͖Δͷ͸໾һΫϥεɺͩͱਏ͍ ত࿨39೥ʢ1964ʣ4݄ʛIMF8৚ࠃʹҠߦ͢Δɿ೔ຊͷ͋ΏΈ ύϒϦοΫυϝΠϯ

Slide 32

Slide 32 text

ࣾ಺ௐ੔໰୊΁ͷղܾࡦ • ໰୊ఆٛϑΣʔζͰߟྀ͓ͯ͘͠ͷ͕Ұ൪ • ͋·Γʹ΋ॏཁ͗͢ΔσʔλΛ࠷ॳ͔Βѻ͏ͷ͸ආ͚Δ • ηϯγςΟϒͳ৘ใؚ͕·ΕΔ΋ͷ • ࣾ಺ͷػີ౓͕͋·Γʹ΋ߴ͍΋ͷ • ͱ͸͍͑ɺInternetͷσʔλ͚ͩͰղܾ͠Α͏ͱ͠ͳ͍ • ݕࡧΤϯδϯ͕ڝ߹ʹͳͬͯ͠·͏

Slide 33

Slide 33 text

3-2. αʔϏεԽ࣌ͷߟྀࣄ߲ଟ͗͢ • ਫ਼౓ݕূΛߦ͏ͨΊͷίʔυ͸Ԛ͘ͳΓ͕ͪ • ͢͹΍͘มߋͰ͖Δɺͷҙຯཱ͕৔ʹΑΓҧ͏ • ΤϯδχΞ : ͙͢ʹมߋΛσϓϩΠͰ͖Δ • σʔλαΠΤϯςΟετ : ͙͢ʹશମΛมߋͰ͖Δ • αʔϏεԽ͠Α͏ͱͨ͠Βɺֶशॲཧ͕ॏ͗ͯ͢ऴΘΒͳ͍ • ͦ΋ͦ΋ɺ඼࣭؅ཧͷ؍఺͕·ͩ੔ཧ͞Ε͍ͯͳ͍

Slide 34

Slide 34 text

͜Εͱ͍ͬͨղܾࡦ͸ͳ͘ɺ೰Έத • ΤϯδχΞ͕AIΛशಘ͢Δʁ • σʔλαΠΤϯςΟετ͕ΤϯδχΞϦϯάΛ਎ʹ͚ͭΔʁ • ۀ຿ௐ੔ΛߦͬͯΏΔ͍ӡ༻Ͱ΋ड͚ೖΕͯ΋Β͏ʁ • ΠϯϑϥଆͰ͋Δఔ౓ٵऩͰ͖ΔΑ͏ʹઃܭ͢Δʁ • CI/CDΛߦ͏ͱͯ͠ςετ͸ʁ؂ࢹͷࢦඪ͸ʁ Image by Convegni_Ancisa from Pixabay

Slide 35

Slide 35 text

ػցֶशϓϩδΣΫτͰ௚໘ͨ͠՝୊ 1.໰୊ఆٛ 2.Ծઆઃఆ 3.։ൃ 4.ޮՌݕূ <-

Slide 36

Slide 36 text

ޮՌݕূϑΣʔζ • ࡞ͬͯΈͨ΋ͷΛ࣮ࡍʹར༻ͯ͠ޮՌ͕ग़Δ͔Ͳ͏͔ݕূ͢Δ • ਫ਼౓͚ͩͰ͸ͳ͘ɺϏδωεతͳՁ஋ΛؚΊͯݕূ • ۤ࿑࿩4-1. ࢥͬͨΑ͏ʹ੒Ռ͕ग़ͳ͍

Slide 37

Slide 37 text

4-1. ࢥͬͨΑ͏ʹ੒Ռ͕ग़ͳ͍ 1.࢖͍ͬͯΔ͚ΕͲ੒Ռ͕ग़ͳ͍ • ։ൃ࣌ͱར༻࣌ͱͰɺσʔλͷ܏޲͕มΘ͍ͬͯΔ 2.ͦ΋ͦ΋ݱ৔͕࢖͍ͬͯͳ͍ • γεςϜ؂ࢹج൫Ͱޡใ๷ࢭͷͨΊʹҟৗݕ஌ػೳΛΦϑ • ࡞ͬͨBotʹ୭΋࿩͔͚͠ͳ͍ Image from ސ٬͕ຊ౰ʹඞཁͩͬͨ΋ͷͱ͸ - χίχίେඦՊ

Slide 38

Slide 38 text

࢖ͬͯ΋Β͏ͨΊʹ • ஍ಓͳܒ໤׆ಈΛ࣮ࢪɺΩϟζϜཧ࿦Λ಄ʹೖΕͯؤுΔ Technology-Adoption-Lifecycle by Craig Chelius CC BY 3.0

Slide 39

Slide 39 text

AIΛ1͍ͭͩ͘͞ • AI͸͖ͪΜͱ࢖͑͹໾ʹཱͭ • ໾ʹཱͭ·Ͱͷ࡞ۀ͸͔͍͍ͬ͜ͱݴ͏ΑΓ͸Ή͠Ζటष͍ • ௻Δ͠ͷAIΛങ͖ͬͯͯσʔλΛ৯ΘͤΕ͹উखʹҭͭͱ͍͏ ΋ͷͰ͸ͳ͍ͷͰɺखΛԚͯ͠গͣͭ͠ਐΊ͍ͯ͘ͷ͕ݱ࣮త Image by Seanbatty from Pixabay