Graph Neural Networks概観

Slide 1

Slide 1 text

ୈ1ճάϥϑษڧձ@KERNEL Graph Neural Networks֓؍ 2019.06.01 fukuגࣜձࣾ ࢁాྋଠ@roy29fuku 1

Slide 2

Slide 2 text

Swampdog ࣗݾ঺հ ࢁాྋଠ@roy29fuku 2013೥ʙ  ɹ౦େ्ҩֶՊͰ໔Ӹݚڀ 2017೥ʙ ɹ޻ֶ෦΁సֶ෦ ɹVRղ๤ֶڭࡐΛେֶͱڞಉ։ൃ 2018೥ ɹfukuגࣜձࣾઃཱɾࢿۚௐୡ ɹ࿦จ෼ੳࣄۀΛ։࢝ ɹ9೥͔͚ͯଔۀͰ͖·ͨ͠ 2

Slide 3

Slide 3 text

# ਪ঑ϓϩτίϧ 1. 8िྸΦεͷϚ΢εAΛ12ඖ༻ҙ 2. 5×105ͷࡉ๔BΛi.v. 3-1. ౤༩܈6ඖʹༀࡎYΛ10μg/kg i.p. 3-2. ରর܈6ඖʹsalineΛi.p. # ධՁ ମԹɾ݂ѹ ࣬ױXʹༀࡎY͕ ޮ͔͘ௐ΂͍ͨ ࣄۀ֓ཁ 3 3 ৽͍͠ݚڀΞΠσΟΞʹରͯ͠ɺ աڈͷ࿦จ͔Βऔಘ࣮ͨ͠ݧ৚݅Λ΋ͱʹ࣮ݧϓϩτίϧΛఏҊ ௐࠪ࣌ؒ΍࣮ݧͷ΍Γ௚͠Λ࡟ݮ

Slide 4

Slide 4 text

ख๏ 4 4 named entity recognition, relation extraction, coreference resolutionΛ༻͍ͯ ࿦จ͔Βϓϩτίϧ΍݁Ռʹؔ͢ΔσʔλΛநग़ɺߏ଄Խ σʔλΛάϥϑߏ଄ʹஔ͖׵͑Δ͜ͱͰϓϩτίϧͷൺֱ΍ੜ੒͕Ͱ͖Δ

Slide 5

Slide 5 text

ख๏ 5 5 named entity recognition, relation extraction, coreference resolutionΛ༻͍ͯ ࿦จ͔Βϓϩτίϧ΍݁Ռʹؔ͢ΔσʔλΛநग़ɺߏ଄Խ σʔλΛάϥϑߏ଄ʹஔ͖׵͑Δ͜ͱͰϓϩτίϧͷൺֱ΍ੜ੒͕Ͱ͖Δ ɹɹɹɹɹɹɹɹɹɹɹɹɹɹɹɹɹ͔Ͳ͏͔֬ೝ͢ΔͨΊʹGNNsษڧ࢝Ί·ͨ͠

Slide 6

Slide 6 text

ख๏ 6 https://speakerdeck.com/ioki/development-of-mrr-at-cookpadtechconf2019 named entity recognition, relation extraction, coreference resolutionΛ༻͍ͯ ࿦จ͔Βϓϩτίϧ΍݁Ռʹؔ͢ΔσʔλΛநग़ɺߏ଄Խ σʔλΛάϥϑߏ଄ʹஔ͖׵͑Δ͜ͱͰϓϩτίϧͷൺֱ΍ੜ੒͕Ͱ͖Δ ɹɹɹɹɹɹɹɹɹɹɹɹɹɹɹɹɹ͔Ͳ͏͔֬ೝ͢ΔͨΊʹGNNsษڧ࢝Ί·ͨ͠ Luan et al., 2018 # ઌߦݚڀ Պֶ࿦จ͔Βknowledge graph ϨγϐΛػցՄಡʹ

Slide 7

Slide 7 text

ख๏ 7 named entity recognition, relation extraction, coreference resolutionΛ༻͍ͯ ࿦จ͔Βϓϩτίϧ΍݁Ռʹؔ͢ΔσʔλΛநग़ɺߏ଄Խ σʔλΛάϥϑߏ଄ʹஔ͖׵͑Δ͜ͱͰϓϩτίϧͷൺֱ΍ੜ੒͕Ͱ͖Δ ɹɹɹɹɹɹɹɹɹɹɹɹɹɹɹɹɹ͔Ͳ͏͔֬ೝ͢ΔͨΊʹGNNsษڧ࢝Ί·ͨ͠ Luan et al., 2018 # ઌߦݚڀ Պֶ࿦จ͔Βknowledge graph ϨγϐΛػցՄಡʹ ϝϯόʔืूத ڵຯ͋Δํ ࿈བྷ͍ͩ͘͞ʂ Twitter: @roy29fuku https://speakerdeck.com/ioki/development-of-mrr-at-cookpadtechconf2019

Slide 8

Slide 8 text

ษڧձͷ໨త 8 8 Graph Neural Networks͕Ͳ͏͍ͬͨྖҬɺ՝୊Λ্ख͘ѻ͑Δ͔஌Γ͍ͨ ɾGNNsʹ͍ͭͯҰॹʹษڧ͠·͠ΐ͏ʂ ɾ֤ྖҬʹ͓͚Δ੒ޭɺࣦഊࣄྫͷڞ༗Λ͠·͠ΐ͏ʂʂ ɾݴޠॲཧֶձͰҟ෼໺ަྲָྀ͕ͦ͠͏͔ͩͬͨΒ ࣍ճҎ߱ͷൃදऀΛืू͍ͯ͠·͢ʂ ɾGraph theoryͷਂ۷Γ ɾGNNsͷਂ۷Γɺ࣮૷ ɾϑϨʔϜϫʔΫ΍ϥΠϒϥϦɺσʔληοτͷ঺հ ɾ෼ੳ݁Ռͷใࠂ ɾͦͷଞͳΜͰ΋

Slide 9

Slide 9 text

ɾGraph ɾGraph Neural Networks ɾNatural Language Processing ɾOther Applications 9 9

Slide 10

Slide 10 text

ɾGraph ɾGraph Neural Networks ɾNatural Language Processing ɾOther Applications 10 10

Slide 11

Slide 11 text

Graphͱ͸ 11 11 ϊʔυʢ௖఺ʣͱΤοδʢลʣͰ ߏ੒͞ΕΔσʔλߏ଄ ϊʔυ: Τοδ:

Slide 12

Slide 12 text

Graphͱ͸ 12 12 ӈͷάϥϑ͸͜Μͳײ͡ʹදݱ ɹ௖఺ू߹V = {1, 2, 3, 4} ɹลू߹E = {{1, 2}, {1, 3}, {2, 3}, {2, 4}, {3, 4}} ɹάϥϑG = (V, E) ผͷݴ͍ํͰఆٛ͢Δͱ ɹE ⊆ [V]2 Λຬͨ͢ ɹG = (V, E) ͷू߹ͷ૊ ΛάϥϑͱݺͿ 1 2 3 4

Slide 13

Slide 13 text

͍ΖΜͳGraph 13 13 ޲͖͕͋Δ ϊʔυ͕multi type Τοδ͕multi type ϧʔϓ͕͋Δ

Slide 14

Slide 14 text

GraphͰදݱͰ͖Δ΋ͷ 14 14 ϊʔυ: ਓʢஉੑ, ঁੑ, …ʣ Τοδ: ਓؒؔ܎ʢ༑ਓ, …ʣ Structured deep models: Deep learning on graphs and beyond karate club

Slide 15

Slide 15 text

GraphͰදݱͰ͖Δ΋ͷ 15 15 ϊʔυ: ݪࢠʢC, H, …ʣ Τοδ: ݁߹ʢ୯݁߹, …ʣ Structured deep models: Deep learning on graphs and beyond

Slide 16

Slide 16 text

GraphͰදݱͰ͖Δ΋ͷ 16 16 ϊʔυ: λϯύΫ࣭ Τοδ: PPI https://academic.oup.com/peds/article/24/9/635/1556325 Structured deep models: Deep learning on graphs and beyond

Slide 17

Slide 17 text

GraphͰදݱͰ͖Δ΋ͷ 17 17 ϊʔυ: ΤϯςΟςΟʢਓ໊, ஍໊, …ʣ Τοδ: ؔ܎ʢॴଐ, ਌ࢠ, …ʣ Structured deep models: Deep learning on graphs and beyond

Slide 18

Slide 18 text

GraphͰදݱͰ͖Δ΋ͷ 18 18 ݱ࣮ͷଟ͘ͷσʔλΛ άϥϑͰදݱͰ͖Δʂ

Slide 19

Slide 19 text

GraphͰղ͖͍ͨ՝୊ 19 19 ׬ᘳͰ͸ͳ͍σʔλͷܽଛ෦෼Λิ׬ͨ͠Γɺ৽͍͠σʔλΛ෼ྨͨ͠Γ ɾϊʔυͷ෼ྨ ɾάϥϑͷ෼ྨʢFYԽֶ෺࣭ͷ෼ྨʣ ɾϦϯΫͷ༧ଌʢϊʔυಉ͕࢜ྡ઀͔൱͔ʣ ɾΤοδͷ෼ྨ Structured deep models: Deep learning on graphs and beyond

Slide 20

Slide 20 text

ɾGraph ɾGraph Neural Networks ɾNatural Language Processing ɾOther Applications 20 20

Slide 21

Slide 21 text

GNNsͷྺ࢙ Structured deep models: Deep learning on graphs and beyond 21

Slide 22

Slide 22 text

GNNsͷྺ࢙ 22 https://www.slideshare.net/takahirokubo7792/graph-attention-network

Slide 23

Slide 23 text

GNNsͷྺ࢙ᶃ ੜ஀ Structured deep models: Deep learning on graphs and beyond GraphΛχϡʔϥϧωοτͰѻͬͨ 23

Slide 24

Slide 24 text

Graph Neueral Networks [Gori+ 2005] 24 24 ֤ϊʔυͷߋ৽ʹ͸ࣗ਎ͱྡ઀ϊʔυΛར༻ Structured deep models: Deep learning on graphs and beyond

Slide 25

Slide 25 text

ne[n]: ϊʔυnͱྡ઀͢Δϊʔυू߹ ln: ϊʔυnͷϥϕϧ xn: ϊʔυnͷঢ়ଶ on: ϊʔυnͷग़ྗ ·ͱΊΔͱ Graph Neueral Networks [Gori+ 2005] 25 25 Fw͕contraction mappingʢॖईࣸ૾ʣͳΒ Fw, Gw͸ղΛ࣋ͪɺunique

Slide 26

Slide 26 text

GNNsͷྺ࢙ᶄ Convolution Structured deep models: Deep learning on graphs and beyond CNNͷ৞ΈࠐΈͷ֓೦Λಋೖͯ͠ܭࢉޮ཰↑ 26 26

Slide 27

Slide 27 text

Convolutional Neural Neworks͓͞Β͍ 27 27 2012: HintonͷAlexNet σΟʔϓϥʔχϯά͕੝Γ্͕͖͔͚ͬͨͬ CNN͸ը૾ͳͲ֨ࢠঢ়ͷσʔλߏ଄ʹద༻Մೳ Structured deep models: Deep learning on graphs and beyond

Slide 28

Slide 28 text

Graph Convolutional Netoworks [Duvenaud+ 2015, Li+ 2016, Scichtkrull+ 2017] 28 28 CNNͷ৞ΈࠐΈͷ֓೦ΛGraphʹద༻ άϥϑߏ଄͸ඇఆܕʢϊʔυͷ਺ɺ֤ϊʔυͷྡ઀ϊʔυ਺ʣ →৞ΈࠐΈͷૢ࡞Λద༻͢ΔͨΊʹ޻෉͕ඞཁ Wu et al., 2019 Euclidean non-Euclidean

Slide 29

Slide 29 text

Graph Convolutional Netoworks [Duvenaud+ 2015, Li+ 2016, Scichtkrull+ 2017] 29 29 CNNͷ৞ΈࠐΈͷ֓೦ΛGraphʹద༻ ΤοδΛॏΈͱͯ͠ѻͬͨ Structured deep models: Deep learning on graphs and beyond

Slide 30

Slide 30 text

Relational Graph Convolutional Netoworks [Scichtkrull+ 2017] 30 30 relationཁૉΛ௥Ճ graph: G = (V, E, R) nodes: vi ∈ V edges: (vi, r, vj) ∈ E relation type: r ∈ R Schlichtkrull et al., 2017

Slide 31

Slide 31 text

Relational Graph Convolutional Netoworks [Scichtkrull+ 2017] 31 31 hi(l): l૚໨ͷϊʔυvi ͷhidden state Ni r: ϊʔυvi ʹରͯؔ͠܎rͰྡ઀͢Δϊʔυू߹ ci,r: ྡ઀͢Δϊʔυͷ਺Λਖ਼نԽɺex. |Nir| ͜ΕΛ֤ϊʔυ͝ͱʹฒྻͰܭࢉɺ࣮ફతʹ͸ܭ ࢉͷޮ཰ͷͨΊʹsparse matrixΛ༻͍Δ ૚ΛॏͶΔ͜ͱͰɺෳ਺ͷϦϨʔγϣϯΛ·͙ͨ ϊʔυಉ࢜ͷؔ܎Λಋ͘ͷʹ໾ཱͭ ͋Δϊʔυʢ੺ʣʹண໨ͨ࣌͠ͷߋ৽ ֤ؔ܎ੑͷೖग़ྗʹؔΘΔྡ઀ϊʔυɾࣗݾϧʔ ϓ΋ܭࢉͯͦ͠ΕͧΕ྘ͷߦྻΛऔಘɺ concatɺReLUͰ෼ྨ Schlichtkrull et al., 2017

Slide 32

Slide 32 text

GNNsͷྺ࢙ᶅ Attention Structured deep models: Deep learning on graphs and beyond attentionΛऔΓೖΕͨ

Slide 33

Slide 33 text

Attention Mechanism͓͞Β͍ 33 33 ೖྗ͝ͱͷॏཁ౓߹͍Λߟྀͨ͠ A Beginner's Guide to Attention Mechanisms and Memory Networks

Slide 34

Slide 34 text

Graph Attention Networks [Monti+ 2017, Hoshen 2017, Veličković+ 2018] 34 34 Ͳͷϊʔυͱͷྡ઀Λॏࢹ͢Δ͔attentionͰදݱͨ͠ GCNͰaij(1/cij)͸ྡ઀ϊʔυ਺ʹґଘ͍ͯͨ͠ Graph Attention NetworksͰ͸ΑΓॏཁͳ΋ͷΛॏΈ෇͚Δ Wu et al., 2019 GCN Graph Attention Networks

Slide 35

Slide 35 text

GNNsͷྺ࢙ᶆ άϥϑͷੜ੒ Structured deep models: Deep learning on graphs and beyond Version 1: Generate graph (or predict new links) between known entities Version 2: Generate graphs from scratch (single embedding vector)

Slide 36

Slide 36 text

GraphͰղ͖͍ͨ՝୊ “ݹయత*”ͳ΋ͷ 36 36 Structured deep models: Deep learning on graphs and beyond ׬ᘳͰ͸ͳ͍σʔλͷܽଛ෦෼Λิ׬ͨ͠Γɺ৽͍͠σʔλΛ෼ྨͨ͠Γ *Kipfᐌ͘classical

Slide 37

Slide 37 text

GraphͰղ͖͍ͨ՝୊ “৽͍͠”΋ͷ 37 37 (SBQI"VUPFODPEFST ྗֶత૬ޓ࡞༻Λߏ଄Λௐ΂Δ ͜ͱͳ͘ɺάϥϑ͔Βਪఆ͢Δ ҼՌؔ܎ͷਪఆ΍ λϯύΫ࣭૬ޓ࡞༻ͷղ໌ʹظ଴ Structured deep models: Deep learning on graphs and beyond

Slide 38

Slide 38 text

GraphͰղ͖͍ͨ՝୊ “৽͍͠”΋ͷ 38 38 (SBQI(FOFSBUJWF/FUXPSLT .PM("/ ෼ࢠΛάϥϑͰදݱ EFTDSJNJOBUPS("/Ͱֶश SFXPSE3-Ͱֶश ෼ࢠͷੜ੒΁ͷԠ༻͕ظ଴ Structured deep models: Deep learning on graphs and beyond

Slide 39

Slide 39 text

ɾGraph ɾGraph Neural Networks ɾNatural Language Processing ɾOther Applications 39 39

Slide 40

Slide 40 text

NLP×Graph 40 40 Zhou et al., 2019

Slide 41

Slide 41 text

ϨγϐΛάϥϑʹམͱ͠ࠐΉ ɹϊʔυ: ࡐྉ ɹΤοδ: खॱ Ԡ༻ͱͯ͠ҎԼͷ࣮ݱΛݟਾ͍͑ͯΔ ɾதؒϊʔυͷݕࡧʢᖱΊͨϕʔίϯʣ ɾෳ਺ϨγϐΛϚʔδ ɾ৽͍͠ௐཧ๏ͷ୳ࡧ 2019/02/27ͷൃදͰ෼ྔͷਖ਼نԽ ࣮ࣾձͷσʔλΛѻ͏͠ΜͲΈʹγϯύγʔ Machine Readable Recipe (recipe) 41 41 https://speakerdeck.com/ioki/development-of-mrr-at-cookpadtechconf2019

Slide 42

Slide 42 text

SCIERC (scientiﬁc protocol) 42 42 Պֶ࿦จʹରͯ͠entities, relations, coreferencesΛΞϊςʔγϣϯͨ͠σʔλ ηοτΛ࡞੒ طଘͷϞσϧ͕ݸผͷλεΫΛύΠϓϥΠϯ తʹܨ͍Ͱͨͷʹର͠ɺmulti-task learning frameworkΛ࣮૷ͨ͠ υϝΠϯݻ༗ͷಛ௃ྔΛ༻͍ͣʹैདྷख๏Λ ্ճΔநग़ਫ਼౓Λ࣮ݱ σʔληοτ ɾ500ຊ෼ͷAbstract ɾentity: 6 types ɾrelation: 7 types Պֶత಺༰ʹಛԽͨ͠৘ใநग़Λ໨ࢦ͢ Luan et al., 2018

Slide 43

Slide 43 text

PaperRobot (knowledge graph) 43 43 աڈͷ࿦จ͔Β knowldge graph*Λੜ੒ ɹnode: entities/concepts ɹedge: relations * άϥϑߏ଄knowledge base ɹʢent1, rel, ent2ʣͷtripletͰදݱ ৽͍͠Պֶతൃݟͱ͸ ৽͍͠node/relationͷൃݟͱଊ͑Δ = incremental work ೖྗ͞ΕͨλΠτϧʹԠͨ֓͡ཁ/݁࿦/ల๬Λࣗಈੜ੒͢Δ ͞Βʹੜ੒͞Εͨ࿦จ͔Β৽͍͠࿦จΛੜ੒͠ɺࣗಈͰແݶʹݚڀΛ͢Δͱ͍͏ߏ૝ 2019/5/20ʹग़ͨϗϠϗϠ࿦จ Wang et al., 2019 SCIERCͷ1st authorͷYi Luan͞Μ͕last author

Slide 44

Slide 44 text

Learning to Represent Programs with Graphs (source code) 44 44 ιʔείʔυͷsemanticͳ෼ੳ NLPతͳΞϓϩʔνͰ͸ಉ͡ม਺΍ؔ ਺͕཭Εͨͱ͜ΖͰ࢖ΘΕͨ࣌ͷґଘ ؔ܎ΛղܾͰ͖ͳ͔ͬͨ Gated Graph Neural NetworksΛར༻ Allamanis et al., 2018 VarNaming: จ຺͔Βม਺Λ໋໊ VarMisuse: ม਺ͷޡ༻Λݕ஌

Slide 45

Slide 45 text

͓͢͢Ίblogهࣄ 45 45 Graph ConvolutionΛࣗવݴޠॲཧʹԠ༻͢Δʢશ7ճʣ ٕज़ಋೖʹ͓͚ΔԾઆݕূͷϓϩηεΛΞ΢τϓοτɺͱͯ΋ษڧʹͳΔ Part7: ߟ࡯ ɾTransformerͷSelf-AttentionͷΑ͏ʹɺࣗવݴޠॲཧʹ͓͍ͯάϥϑతͳػߏ͕༗༻ͳ৔ ߹͸͋Δɻ͔͠͠ɺͦͷ͜ͱͱGraph Convolution͕༗ޮͳ͜ͱ͸౳ՁͰ͸ͳ͍ɻ ɾGraph Convolution͕༗ޮͳͷ͸ɺϊʔυ෼ྨ/άϥϑߏ଄෼ྨͷλεΫʹམͱ͠ࠐΊΔ έʔεɻจ຺৘ใΛ֫ಘ͍ͨ͠ͱ͍͏ϞνϕʔγϣϯͱɺGraph ConvolutionͷಘҙྖҬͱ ͸͋·Γ߹க͠ͳ͍ɻ ɾGraph ConvolutionΛࣗવݴޠॲཧͰ࢖͏ͳΒ͹ɺʮେن໛͔ͭHeterogeneousͳάϥϑ Ͱɺϊʔυ෼ྨͷ໰୊ʹؼணͰ͖Δʯέʔε͕ద͍ͯ͠ΔͱࢥΘΕΔɻ

Slide 46

Slide 46 text

ɾGraph ɾGraph Neural Networks ɾNatural Language Processing ɾOther Applications 46 46

Slide 47

Slide 47 text

MoleculeNet (chemical) 47 47 molecular machine learningͷϕϯ νϚʔΫͱͳΔσʔληοτ ྔࢠྗֶɾ෺ཧԽֶɾੜ෺෺ཧֶɾ ੜཧֶͷ؍఺͔ΒूΊΒΕͨ 700,000ͷԽ߹෺σʔλͰߏ੒ ෳ਺ͷGNNs͕σϑΥϧτͰ࢖͑Δ drug discoveryϑϨʔϜϫʔΫͷ DeepChem಺ʹͯσʔληοτͱ͠ ͯ࢖ΘΕ͍ͯΔ Wu et al., 2018

Slide 48

Slide 48 text

BrainNetCNN (neuroscience) 48 48 ίωΫτʔϜʢਆܦճ࿏ʣΛ NNsʹಥͬࠐΜͩ ۩ମతʹ͸ૣ࢈ͷ༮ࣇͷ֦ࢄςϯιϧը૾ ʢDTIʣΛ࢖ͬͯೝ஌ೳྗͱӡಈೳྗͷ༧ ଌʹ༻͍ͨ Τοδʹྔతͳม਺Λ༩͑ɺਆܦͷ઀ଓͷ ڧ͞Λදݱ edge-to-edge, edge-to-node and node-to- graphͳ৞ΈࠐΈϑΟϧλʔΛར༻ Kawahara et al., 2017

Slide 49

Slide 49 text

Other Applications 49 49 Structured deep models: Deep learning on graphs and beyond

Slide 50

Slide 50 text

Other Applications 50 50 Zhou et al., 2019

Slide 51

Slide 51 text

Datasets 51 51 Wu et al., 2019

Slide 52

Slide 52 text

࣮ࡍGNNsͬͯ࢖͑Δͷʁ 52 52

Slide 53

Slide 53 text

1-dimensional Weisfeiler-Leman graphͱಉ౳ https://www.youtube.com/watch?v=DLjmQmfRlis

Slide 54

Slide 54 text

ϩʔύεϑΟϧλʔͱಉ౳ 54 54 https://github.com/arXivTimes/arXivTimes/issues/1232 https://twitter.com/shion_honda/status/1132146301024391169

Slide 55

Slide 55 text

References 55 55

Slide 56

Slide 56 text

56 (SBQIཧ࿦ ˒͞ΜΦεεϝ%BOJFM4QJFMNBOߨٛࢿྉ ˒͞ΜΦεεϝ%BOJFM4QJFMNBOߨٛಈը (//T ˒ࢁా͓͢͢Ί,JQGͷϖʔδࢿྉ4USVDUVSFEEFFQNPEFMT%FFQMFBSOJOHPOHSBQITBOECFZPOE ˒ࢁా͓͢͢ΊSFWJFX"$PNQSFIFOTJWF4VSWFZPO(SBQI/FVSBM/FUXPSLT SFWJFX(SBQI/FVSBM/FUXPSLT"3FWJFXPG.FUIPETBOE"QQMJDBUJPOT ,JQGͷ·ͱΊ(3"1)$0/70-65*0/"-/&5803,4 JDPYGPHHSBQIDPOWPMVUJPOOMQ (//T"/FX.PEFMGPS-FBSOJOHJO(SBQI%PNBJOT 3($/T.PEFMJOH3FMBUJPOBM%BUBXJUI(SBQI$POWPMVUJPOBM/FUXPSLT .PEFMJOH3FMBUJPOBM%BUBXJUI(SBQI$POWPMVUJPOBM/FUXPSLTͷ·ͱΊ (//·ͱΊ d (SBQI"UUFOUJPO/FUXPSL References 1/2

Slide 57

Slide 57 text

57 /-1 .33 .BDIJOF3FBEBCMF3FDJQF ͷ։ൃ%FWFMPQNFOUPG.33!$PPLQBE5FDI$POG 4$*&3$ 1BQFS3PCPU -FBSOJOHUP3FQSFTFOU1SPHSBNTXJUI(SBQIT (SBQI$POWPMVUJPOͰࣗવݴޠॲཧΛߦ͏ 0UIFS"QQMJDBUJPOT .PMFDVMF/FU #SBJO/FU$// ࣮ࡍ(//Tͬͯ࢖͑Δͷʁ )PX1PXFSGVMBSF(SBQI/FVSBM/FUXPSLT 8FJTGFJMFSBOE-FNBO(P/FVSBM)JHIFSPSEFS(SBQI/FVSBM/FUXPSLT 3FWJTJUJOH(SBQI/FVSBM/FUXPSLT"MM8F)BWFJT-PX1BTT'JMUFST (SBQI/FVSBM/FUXPSL3FUSJFWB5&$)#-0( 5XJUUFS!TIJPO@IPOEB IUUQTHJUIVCDPNBS9JW5JNFTBS9JW5JNFTJTTVFT References 2/2

Slide 58

Slide 58 text

͍͞͝ʹ 58 58

Slide 59

Slide 59 text

59 59 ࣍ճҎ߱ͷൃදऀΛืू͍ͯ͠·͢ʂ ɾGraph theoryͷਂ۷Γ ɾGNNsͷαʔϕΠ, ࣮૷ ɾϑϨʔϜϫʔΫ, ϥΠϒϥϦ, σʔληοτͷ঺հ ɾ෼ੳ݁Ռͷใࠂ ɾͦͷଞͳΜͰ΋ ख఻͍͚ͬͯͨͩΔํ͸SlackͳΓTwitterͳΓͰ ࢁాʢroy29fukuʣ·Ͱ࿈བྷ͍ͩ͘͞ʂ ཁ๬͕͋Ε͹࣍ճҎ߱ଓ͖·͢