Tweet
Tweet

Slide 1

Slide 1 text

/ 166 1 堣唒㷕统ה堣唒涪鋅 ر٦ة⚥䗰㘗ך⻉㷕٥勞俱猰㷕ך侄鎮הֿ׸ַ׵ 戣䊛♧㷕 ichigaku.takigawa@riken.jp 椚⻉㷕灇瑔䨽ꬠ倜濼腉窟さ灇瑔إٝة٦ J14稢脄鸬䵿⼔㷕涸ٔأؙ㔐鼘ث٦ي 2021䎃10剢26傈

Slide 2

Slide 2 text

/ 166 㼔Ꟍ堣唒㷕统ה堣唒涪鋅 խխխ暴חꨄ侔圓鸡׾⠵ֲ堣唒㷕统ر٦ة⚥䗰涸ז荈搫猰㷕灇瑔 խխխ植㖈ך⚺噟⹡䎌稢脄欰暟㷕 椚灇 ⻉㷕 ⻌㣐 荈䊹稱➜戣䊛♧㷕 ׋ָֹ׻ְ׍ָֻ 2 䎃⻌㣐 ։ 䎃❨㣐 ։ 䎃⻌㣐 ։ 䊨㷕灇瑔猰ءأذي䞔㜠䊨㷕㼔余⽆㡦铬玎⥜✪ ⸋寸㹀⥋〾彁ⴓꨄך-ظٕي剑㼭鍑ך椚锷ⴓ匿 ⻉㷕灇瑔䨽غ؎ؔ؎ٝؿؓوذ؍ؙأإٝة٦ 讒㷕灇瑔猰⼔讒ⶼ䧭䞔㜠猰㷕㼔余 Ⱟ⹡ 䞔㜠猰㷕灇瑔猰䞔㜠椚䊨㷕㼔余 +45ָֹֽׁ勞俱؎ٝؿؓوذ؍ؙأ걄㚖 Ⱟ⹡ 䎃椚灇 ❨鿪 ։ "*1إٝة٦J14稢脄鸬䵿⼔㷕涸ٔأؙ㔐鼘ث٦ي ⻌㣐⻉㷕⿾䘔ⶼ䧭灇瑔䬿挿 ؙٗأ،ه؎ٝز ⸔侄 Ⲥ侄䱇 灇瑔㆞

Slide 3

Slide 3 text

/ 166 ꨄ侔圓鸡׾⠵ֲ堣唒㷕统 3 ܾఆ໦ɾܾఆDAG χϡʔϥϧωοτϫʔΫ ֬཰తϓϩάϥϛϯά ˖ 㼎韋ָչꨄ侔圓鸡պ׾䭯א ˖ ٌرָٕչꨄ侔圓鸡պ׾䭯א ˖ 㼎韋ךꟼ⤘ָչꨄ侔圓鸡պ׾䭯א ꧊さծ锷椚ծꟼ⤘ծ穈さׇծ禸⴨ծ加ծؚٓؿծ➿侧禸ծ鎉铂ծ˘ CH 3 N N H N H H 3 C N

Slide 4

Slide 4 text

/ 166 ➙傈ךذ٦و 4 ˖ 荈䊹稱➜ 堣唒㷕统ה荈搫猰㷕ך㞮歲 ˖ 堣唒㷕统הכ倜׃ְفؚٗٓىؚٝך倯岀 ˖ 堣唒㷕统㾊כ♧⡤⡦ָ嚂׃ְךַ ˖ ⴓ㶨ך邌植ה堣唒㷕统 ˖ ؚٖ؎نحؙأ剑黝⻉怴糊䌓秛 锷椚䱿锷ה窟鎘涸✮庠 ך窟さ ˖ 荈搫猰㷕灇瑔ד堣唒㷕统׾⢪ֲֶהׅ׷ה䗳׆עאַ׷劤䔲חꨇ׃ְ㉏겗 ˖ 5IF5XP$VMUVSFTر٦ةٌرؚٔٝה✮庠،ٕ؞ٔؤي ˖ ✮庠ַ椚鍑ַ3BTIPNPO⸬卓 6OEFSTQFDJDBUJPO 鍑ꅸ㢳圫䚍 ˖ ➂꟦ך钠濼غ؎،أח歋勻ׅ׷㉏겗⟎铡ծ㣟侁ծ䧭⸆غ؎،أծFUD ˖ 堣唒㷕统ַ׵堣唒涪鋅פ ˖ չ涪鋅պչ椚鍑պך麣瘡כさ椚⻉דֹ׷ךַ荈⹛⻉דֹ׷ךַ

Slide 5

Slide 5 text

/ 166 鯄耵䎃剢傈։ 5 ⻌嵲麣㣐㷕䞔㜠猰㷕灇瑔猰ך灇瑔㹓׾DMPTF׃♴鎸穈籼ך չؙٗأ،ه؎ٝزًٝزպפ ˖ 椚⻉㷕灇瑔䨽 ꬠ倜濼腉窟さ灇瑔إٝة٦ "*1 J14稢脄鸬䵿⼔㷕涸ٔأؙ㔐鼘ث٦ي灇瑔㆞ ˖ ⻌嵲麣㣐㷕 ⻉㷕⿾䘔ⶼ䧭灇瑔䬿挿 81**$3F%% 暴⟣Ⲥ侄䱇

Slide 6

Slide 6 text

/ 166 俑猰満⚅歲زحفٖكٕ䬿挿䕎䧭فؚٗٓي 81* 6 https://www.mext.go.jp/a_menu/kagaku/toplevel/

Slide 7

Slide 7 text

/ 166 ⚅歲زحفٖكٕ䬿挿䕎䧭فؚٗٓي 81* ה䞔㜠猰㷕 7 ر٦ة٥䞔㜠猰㷕חꅾז׷ 81*䬿挿כ✳䬿挿ך׫ صُ٦ٗ؎ٝذٔآؑٝأ㕂ꥷ灇瑔堣圓

Slide 8

Slide 8 text

/ 166 ⻌嵲麣㣐㷕⻉㷕⿾䘔ⶼ䧭灇瑔䬿挿 81**$3F%% 8 ⻉㷕⿾䘔ה䞔㜠猰㷕

Slide 9

Slide 9 text

/ 166 ⻌嵲麣㣐㷕⻉㷕⿾䘔ⶼ䧭灇瑔䬿挿 81**$3F%% 9 https://www.icredd.hokudai.ac.jp

Slide 10

Slide 10 text

/ 166 ظ٦كٕ⻉㷕颣 10

Slide 11

Slide 11 text

/ 166 https://www.youtube.com/watch?v=clvA49BobsI 11

Slide 12

Slide 12 text

/ 166 椚⻉㷕灇瑔䨽ꬠ倜濼腉窟さ灇瑔إٝة٦ 12 https://aip.riken.jp/

Slide 13

Slide 13 text

/ 166 椚⻉㷕灇瑔䨽ꬠ倜濼腉窟さ灇瑔إٝة٦ 13 東京駅 理研AIP 皇居

Slide 14

Slide 14 text

/ 166 椚⻉㷕灇瑔䨽ꬠ倜濼腉窟さ灇瑔إٝة٦ 14 $03&%0傈劤堀'

Slide 15

Slide 15 text

/ 166 15 https://www.kobe.riken.jp/about/map/keihanna/ ⹶⹡㖑❨ꢻ㣽㖑⼒ ❨鿪䏍湱嚂龾礵螟歕

Slide 16

Slide 16 text

/ 166 ⹶⹡㖑❨ꢻ㣽㖑⼒ ❨鿪䏍湱嚂龾礵螟歕 16

Slide 17

Slide 17 text

/ 166 ⹶⹡㖑❨ꢻ㣽㖑⼒ ❨鿪䏍湱嚂龾礵螟歕 17 ࠃཱࠃձਤॻؗؔ੢ؗ

Slide 18

Slide 18 text

/ 166 㕂ꥷꨵ孡鸐⥋㛇燉䪮遭灇瑔䨽"53 18 瀖랲暴灇ך ،ٝسٗ؎س չؒٔؕպָꔨ䏟 㹋暟כ乆䕦犜姺 ˖ 椚⻉㷕灇瑔䨽 ꬠ倜濼腉窟さ灇瑔إٝة٦"*1 ؖ٦ر؍،ٝٗنحزفٗآؙؑز(31 ˖ "53膷䞔㜠鸐⥋筨さ灇瑔䨽 膷䞔㜠灇瑔䨽 钠濼堣圓灇瑔䨽 膷䞔㜠鍑匿灇瑔䨽 https://www.atr.jp ˖ 帾㾴؎ٝةؙٓءّٝ ؎ٝةؙٓءّٝ䪮遭غؙٝ ؎ٝةؙٓءّٝ猰 㷕灇瑔䨽 瀖랲嵞暴ⴽ灇瑔䨽 蠗歊秀⽆暴ⴽ灇瑔䨽 ˖ 搀简٥鸐⥋ 黝䘔؝ىُص؛٦ءّٝ灇瑔䨽 岚⹛䊨㷕灇瑔䨽 ˖ 欰ㄏ猰㷕 ⡟谏⻟䗗暴ⴽ灇瑔䨽

Slide 19

Slide 19 text

/ 166 19 ˖ 膷䞔㜠灇瑔䨽 $/4 ˖ 钠濼堣圓灇瑔䨽 $.$ ˖ 膷䞔㜠鍑匿灇瑔䨽 /*" ˖ 鎘皾膷؎ً٦آؚٝ灇瑔㹓 $#* ͑椚灇"*1鎘皾膷ت؎شىؙأث٦ي 㿊♴5 ˖ ⹛涸膷؎ً٦آؚٝ灇瑔㹓 %#* ͑椚灇"*1膷䞔㜠窟さ鍑匿ث٦ي 䊛ꑚ5 ˖ 膷䞔㜠鸐⥋筨さ灇瑔䨽 㕂ꥷꨵ孡鸐⥋㛇燉䪮遭灇瑔䨽"53

Slide 20

Slide 20 text

/ 166 椚灇"*1!"53 20 ˖ ꣇拄猰㷕ث٦ي ♳歊⥜⸆ ˖ 膷䞔㜠窟さ鍑匿ث٦ي 䊛ꑚ♧儙 ˖ 鎘皾膷ت؎شىؙأث٦ي 㿊♴㸾➂ ˖ J14稢脄鸬䵿⼔㷕涸ٔأؙ㔐鼘ث٦ي ♳歊⥜⸆ 椚灇"*1ה❨㣐J14稢脄灇ך鸬䵿ٓن

Slide 21

Slide 21 text

/ 166 植㖈ךꟼ䗰 21 ˖ ⻌嵲麣㣐㷕⻉㷕⿾䘔ⶼ䧭灇瑔䬿挿 81**$3F%% ˖ 椚⻉㷕灇瑔䨽ꬠ倜濼腉窟さ灇瑔إٝة٦ "*1 ˖ ꨄ侔圓鸡٥穈さׇ圓鸡׾⠵ֲ堣唒㷕统 ˖ 䎌稢脄欰暟㷕ך׋׭ך堣唒㷕统 稢脄歗⫷帾㾴㷕统 ˖ 倜׃ְ،ٕ؞ٔؤي٥剑黝⻉ך㹀䒭⻉ה実鍑岀 ˖ 鷲麓㘗ꨵ㶨겥䗍ꖎ堣唒㷕统ח״׷⹛涸錁㻊 ˖ 堣唒㷕统ך㹋騧灇瑔 ˖ ⻉㷕⿾䘔ךرؠ؎ٝה涪鋅ך׋׭ך堣唒㷕统 ˖ ⴓ㶨ךؚٓؿ邌植ך㷕统ה欰䧭 ˖ ꆀ㶨⻉㷕鎘皾堣唒㷕统ך輐さ ˖ 堣唒涪鋅䱱稊ծ㹋꿀鎘歗ծ濼陎涪鋅

Slide 22

Slide 22 text

/ 166 ➙傈ךذ٦و 22 ˖ 荈䊹稱➜ 堣唒㷕统ה荈搫猰㷕ך㞮歲 ˖ 堣唒㷕统הכ倜׃ְفؚٗٓىؚٝך倯岀 ˖ 堣唒㷕统㾊כ♧⡤⡦ָ嚂׃ְךַ ˖ ⴓ㶨ך邌植ה堣唒㷕统 ˖ ؚٖ؎نحؙأ剑黝⻉ 怴糊䌓秛 锷椚㷕ה窟鎘㷕ך輐さ ˖ 荈搫猰㷕灇瑔ד堣唒㷕统׾⢪ֲֶהׅ׷ה䗳׆עאַ׷劤䔲חꨇ׃ְ㉏겗 ˖ ر٦ةٌرؚٔٝה✮庠،ٕ؞ٔؤي 5IF5XP$VMUVSFT ˖ ✮庠ַ椚鍑ַ3BTIPNPO⸬卓 6OEFSTQFDJDBUJPO 鍑ꅸ㢳圫䚍 ˖ ➂꟦ך钠濼غ؎،أח歋勻ׅ׷㉏겗⟎铡ծ㣟侁ծ䧭⸆غ؎،أծFUD ˖ 堣唒㷕统ַ׵堣唒涪鋅פ ˖ չ涪鋅պכさ椚⻉דֹ׷ךַׁ׵ח荈⹛⻉דֹ׷ךַ

Slide 23

Slide 23 text

/ 166 堣唒㷕统הכ倜䩛ך ꧟ז فؚٗٓىؚٝך倯岀 23 ؝ٝؾُ٦ةفؚٗٓي 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 ♷ִ׵׸׋㣐ꆀךⰅ⳿⸂ך鋅劤⢽׾ⱄ植דֹ׷״ֲזⰅ⸂ַ׵⳿⸂פ ך㢌䳔فؚٗٓي׾ꬊ僇爙涸ח欰䧭ׅ׷׋׭ך害欽涸倯岀 ♧菙暟⡤钠陎 갈㡮钠陎 堣唒缺鏬 馄鍑⫷؎ً٦آؚٝ ⴓ㶨ך嫩䚍✮庠 ˑ֮׶ָהֲ˒ J’aime la musique I love music CH3 N H 3 C H N S N O CH3 N OH 1.394

Slide 24

Slide 24 text

/ 166 堣唒㷕统הכ倜䩛ך ꧟ז فؚٗٓىؚٝך倯岀 24 ⠗窟涸זفؚٗٓىؚٝ 怴糊涸ծSBUJPOBM ءىُٖ٦ءّٝ 鎘皾ך׋׭ךٗآحؙכ ׅץג➂꟦ָ罋ִ׷ Ⰵ⸂ ⳿⸂ 倜׃ְفؚٗٓىؚٝ 䌓秛涸ծFNQJSJDBM 堣唒㷕统 Ⰵ⳿⸂ךꟼ⤘כ״ֻⴓַ ׵זְךד镘׭׷ Ⰵ⸂ ⳿⸂ չػًٓةד䮙⹛׾荈歋ח㢌ִ׵׸׷害欽䕎պד ꧵䕎׾欽䠐׃ծ׋ֻׁ׿ךⰅ⳿⸂ך鋅劤⢽׾♷ִג 鋅劤⢽׾ⱄ植ׅ׷״ֲػًٓةך⦼׾锃侭ׅ׷

Slide 25

Slide 25 text

/ 166 堣唒㷕统剣ꣲך挿פךꟼ侧ؿ؍حذ؍ؚٝ 25 x1 x2 y p1 p2 p3 p5 p4 Variable 1 Variable 2 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 x1 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 x2 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 x1 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 x2 x1 x2 y ML 5 params Random Forest Neural Networks SVR Kernel Ridge

Slide 26

Slide 26 text

/ 166 ⴓ겲 DMBTTJDBUJPO BTؿ؍حذ؍ؚٝ 26

Slide 27

Slide 27 text

/ 166 ⴓ겲 DMBTTJDBUJPO BTؿ؍حذ؍ؚٝ 26

Slide 28

Slide 28 text

/ 166 ⴓ겲 DMBTTJDBUJPO BTؿ؍حذ؍ؚٝ 26

Slide 29

Slide 29 text

/ 166 ⴓ겲 DMBTTJDBUJPO BTؿ؍حذ؍ؚٝ 26 ך然桦⦼׾⳿⸂ QSFEJDU@QSPCB P(class=red) P(class=blue) = 1 - P(class=red)

Slide 30

Slide 30 text

/ 166 ⴓ겲 DMBTTJDBUJPO BTؿ؍حذ؍ؚٝ 26 Random Forest Gaussian Process Classifier Logistic Regression ך然桦⦼׾⳿⸂ QSFEJDU@QSPCB P(class=red) P(class=blue) = 1 - P(class=red)

Slide 31

Slide 31 text

/ 166 #PSJOH"* BLB.BDIJOF-FBSOJOH 27 https://www.forbes.com/sites/forbestechcouncil/2020/02/19/ in-praise-of-boring-ai-a-k-a-machine-learning/ ʜ “Let’s face it: So far, the artificial intelligence plastered all over PowerPoint slides hasn’t lived up to its hype.” The AI frenzy: hope & hype

Slide 32

Slide 32 text

/ 166 #PSJOH"* BLB.BDIJOF-FBSOJOH 27 From AAAI-20 Oxford-Style Debate https://www.forbes.com/sites/forbestechcouncil/2020/02/19/ in-praise-of-boring-ai-a-k-a-machine-learning/ ʜ “Let’s face it: So far, the artificial intelligence plastered all over PowerPoint slides hasn’t lived up to its hype.” The AI frenzy: hope & hype

Slide 33

Slide 33 text

/ 166 ꟼ侧ؿ؍حذ؍ؚٝה׃גך堣唒㷕统 28 Prediction Input variables Classifier or Regressor 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 x1 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 x2 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 x3 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 . . . Function model 垥彊涸ז堣唒㷕统ٌرٕ

Slide 34

Slide 34 text

/ 166 邌植㷕统 葺ְ悵㖈暴䗙ꆀךر٦ةַ׵ך䬄⳿ 29 Prediction Input variables Function model AAAChnichVG7TgJBFD2sL8QHqI2JDZFgrMhAVIwV0caShzwSJGR3HXHDvrK7EJH4Aya2UlhpYmH8AD/Axh+w4BOMJSY2Fl6WTYwS8W5m58yZe+6cmSuZqmI7jHV9wtj4xOSUfzowMzs3HwwtLBZso2HJPC8bqmGVJNHmqqLzvKM4Ki+ZFhc1SeVFqb7X3y82uWUrhn7gtExe0cSarhwrsugQlTutJqqhCIsxN8LDIO6BCLxIG6FHHOIIBmQ0oIFDh0NYhQibvjLiYDCJq6BNnEVIcfc5zhEgbYOyOGWIxNbpX6NV2WN1Wvdr2q5aplNUGhYpw4iyF3bPeuyZPbBX9vlnrbZbo++lRbM00HKzGrxYzn38q9JodnDyrRrp2cExtl2vCnk3XaZ/C3mgb551ermdbLS9xm7ZG/m/YV32RDfQm+/yXYZnr0f4kcgLvRg1KP67HcOgkIjFt2KJzEYkteu1yo8VrGKd+pFECvtII0/1a7jEFTqCX4gJm0JykCr4PM0SfoSQ+gJWLpCb x2 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 x3 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 . . . Latent variables Learnable variable transformation Representation learning Classifier or Regressor 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 x1 剑鵚ך帾㾴㷕统דכ㔐䌓٥ⴓ겲ך⵸ח葺ְ悵㖈㢌侧邌植פך㢌䳔׾遤ֲ אךـٗحؙךさ䧭׾ꟼ侧ؿ؍حذ؍ؚٝה׃ג♧孡鸐顐ד剑黝⻉ׅ׷

Slide 35

Slide 35 text

/ 166 邌植㷕统 葺ְ悵㖈暴䗙ꆀךر٦ةַ׵ך䬄⳿ 29 Prediction Input variables Function model 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 x2 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 x3 AAACiXichVG7SgNBFD2ur/hM1EawEYNiFWZFNKQKprGMj0TBBNndTHR0X+xOFmLwB6zsRK0ULMQP8ANs/AELP0EsFWwsvNksiAbjXWbnzJl77pyZq7um8CVjz11Kd09vX39sYHBoeGQ0nhgbL/pOzTN4wXBMx9vWNZ+bwuYFKaTJt12Pa5Zu8i39MNfc3wq45wvH3pR1l5ctbc8WVWFokqhiKag40t9NJFmKhTHdDtQIJBFF3knco4QKHBiowQKHDUnYhAafvh2oYHCJK6NBnEdIhPscxxgkbY2yOGVoxB7Sf49WOxFr07pZ0w/VBp1i0vBIOY1Z9sRu2Rt7ZHfshX3+WasR1mh6qdOst7Tc3Y2fTG58/KuyaJbY/1Z19CxRRTr0Ksi7GzLNWxgtfXB09raRWZ9tzLFr9kr+r9gze6Ab2MG7cbPG1y87+NHJC70YNUj93Y52UFxIqUuphbXFZHYlalUMU5jBPPVjGVmsIo8C1T/AKc5xoQwpqpJWMq1UpSvSTOBHKLkvAi+SPA== . . . Latent variables Learnable variable transformation Representation learning Classifier or Regressor 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 x1 剑鵚ך帾㾴㷕统דכ㔐䌓٥ⴓ겲ך⵸ח葺ְ悵㖈㢌侧邌植פך㢌䳔׾遤ֲ אךـٗحؙךさ䧭׾ꟼ侧ؿ؍حذ؍ؚٝה׃ג♧孡鸐顐ד剑黝⻉ׅ׷

Slide 36

Slide 36 text

/ 166 邌植㷕统 葺ְ悵㖈暴䗙ꆀךر٦ةַ׵ך䬄⳿ 29 Prediction Input variables Function model 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 x2 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 x3 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 . . . Latent variables Learnable variable transformation Representation learning Classifier or Regressor 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 x1 剑鵚ך帾㾴㷕统דכ㔐䌓٥ⴓ겲ך⵸ח葺ְ悵㖈㢌侧邌植פך㢌䳔׾遤ֲ אךـٗحؙךさ䧭׾ꟼ侧ؿ؍حذ؍ؚٝה׃ג♧孡鸐顐ד剑黝⻉ׅ׷

Slide 37

Slide 37 text

/ 166 邌植㷕统 葺ְ悵㖈暴䗙ꆀךر٦ةַ׵ך䬄⳿ 29 Prediction Input variables Function model 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 x2 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 x3 AAACiXichVG7SgNBFD2ur/hM1EawEYNiFWZFNKQKprGMj0TBBNndTHR0X+xOFmLwB6zsRK0ULMQP8ANs/AELP0EsFWwsvNksiAbjXWbnzJl77pyZq7um8CVjz11Kd09vX39sYHBoeGQ0nhgbL/pOzTN4wXBMx9vWNZ+bwuYFKaTJt12Pa5Zu8i39MNfc3wq45wvH3pR1l5ctbc8WVWFokqhiKag40t9NJFmKhTHdDtQIJBFF3knco4QKHBiowQKHDUnYhAafvh2oYHCJK6NBnEdIhPscxxgkbY2yOGVoxB7Sf49WOxFr07pZ0w/VBp1i0vBIOY1Z9sRu2Rt7ZHfshX3+WasR1mh6qdOst7Tc3Y2fTG58/KuyaJbY/1Z19CxRRTr0Ksi7GzLNWxgtfXB09raRWZ9tzLFr9kr+r9gze6Ab2MG7cbPG1y87+NHJC70YNUj93Y52UFxIqUuphbXFZHYlalUMU5jBPPVjGVmsIo8C1T/AKc5xoQwpqpJWMq1UpSvSTOBHKLkvAi+SPA== . . . Latent variables Learnable variable transformation Representation learning Classifier or Regressor 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 x1 剑鵚ך帾㾴㷕统דכ㔐䌓٥ⴓ겲ך⵸ח葺ְ悵㖈㢌侧邌植פך㢌䳔׾遤ֲ אךـٗحؙךさ䧭׾ꟼ侧ؿ؍حذ؍ؚٝה׃ג♧孡鸐顐ד剑黝⻉ׅ׷

Slide 38

Slide 38 text

/ 166 邌植㷕统 葺ְ悵㖈暴䗙ꆀךر٦ةַ׵ך䬄⳿ 29 Prediction Input variables Function model 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 x2 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 x3 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 . . . Latent variables Learnable variable transformation Representation learning Classifier or Regressor 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 x1 剑鵚ך帾㾴㷕统דכ㔐䌓٥ⴓ겲ך⵸ח葺ְ悵㖈㢌侧邌植פך㢌䳔׾遤ֲ אךـٗحؙךさ䧭׾ꟼ侧ؿ؍حذ؍ؚٝה׃ג♧孡鸐顐ד剑黝⻉ׅ׷

Slide 39

Slide 39 text

/ 166 邌植㷕统 葺ְ悵㖈暴䗙ꆀךر٦ةַ׵ך䬄⳿ 29 Prediction Input variables Function model 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 x2 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 x3 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 . . . Latent variables Learnable variable transformation Representation learning Classifier or Regressor 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 x1 剑鵚ך帾㾴㷕统דכ㔐䌓٥ⴓ겲ך⵸ח葺ְ悵㖈㢌侧邌植פך㢌䳔׾遤ֲ אךـٗحؙךさ䧭׾ꟼ侧ؿ؍حذ؍ؚٝה׃ג♧孡鸐顐ד剑黝⻉ׅ׷

Slide 40

Slide 40 text

/ 166 邌植㷕统 葺ְ悵㖈暴䗙ꆀךر٦ةַ׵ך䬄⳿ 29 Prediction Input variables Function model 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 x2 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 x3 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 . . . Latent variables Learnable variable transformation Representation learning Classifier or Regressor Linear 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 x1 Simple model is enough when we have good features. 剑鵚ך帾㾴㷕统דכ㔐䌓٥ⴓ겲ך⵸ח葺ְ悵㖈㢌侧邌植פך㢌䳔׾遤ֲ אךـٗحؙךさ䧭׾ꟼ侧ؿ؍حذ؍ؚٝה׃ג♧孡鸐顐ד剑黝⻉ׅ׷

Slide 41

Slide 41 text

/ 166 邌植㷕统 葺ְ悵㖈暴䗙ꆀךر٦ةַ׵ך䬄⳿ 30 https://colah.github.io/posts/2014-03-NN-Manifolds-Topology/ 简䕎ⴓꨄ〳腉ז邌植פך㢌䳔׾㷕统 ֿךةأؙכ㹋騧涸חכ⣛搫׬׆ַ׃ֻ ꟦麩ִ׷ה⯋״׶ꃎֻז׶ֲ׷

Slide 42

Slide 42 text

/ 166 邌植㷕统 葺ְ悵㖈暴䗙ꆀךر٦ةַ׵ך䬄⳿ 30 https://colah.github.io/posts/2014-03-NN-Manifolds-Topology/ 简䕎ⴓꨄ〳腉ז邌植פך㢌䳔׾㷕统 ֿךةأؙכ㹋騧涸חכ⣛搫׬׆ַ׃ֻ ꟦麩ִ׷ה⯋״׶ꃎֻז׶ֲ׷

Slide 43

Slide 43 text

/ 166 31 https://towardsdatascience.com/a-comprehensive-guide-to-convolutional-neural-networks-the-eli5-way-3bd2b1164a53 ... ... 葺ְ悵㖈㢌侧邌植פך㢌䳔׾ر٦ةַ׵㷕统 ةأؙ׀הח㔐䌓٥ⴓ겲 ֿך鿇ⴓכ✼ְח湱ꟼך֮׷醱侧ך ⴽةأؙדⵃ欽דֹ׷ 鯄獳㷕统 邌植㷕统 葺ְ悵㖈暴䗙ꆀ ךⴽך♴崧ةأؙפך鯄獳

Slide 44

Slide 44 text

/ 166 32 ... ... 葺ְ悵㖈㢌侧邌植פך㢌䳔׾ر٦ةַ׵㷕统 ةأؙ׀הח㔐䌓٥ⴓ겲 Ԩ ر٦ة挿׾Ⰵ⸂邌植דכזֻ悵㖈㢌侧邌植חְֶגⰻ䯏ׅ׷ֿהחז׷կ Ԩ Ⱏ鸐ךչ葺ְ悵㖈㢌侧邌植պ׾䭯אةأؙדչ邌植㷕统ـٗحؙպ׌ֽ׾㷕统 דֹ׷〳腉䚍׾䭯אկ 㣐鋉垷ر٦ةדך✲⵸㷕统̔㼭鋉垷⢽פ鯄獳㷕统 邌植㷕统 葺ְ悵㖈暴䗙ꆀ ךⴽך♴崧ةأؙפך鯄獳

Slide 45

Slide 45 text

/ 166 堣唒㷕统ך植➿涸ז⩎꬗ 33 ˖ ٌرٕػًٓة侧ָה׿ד׮זֻ㢳ְ ˖ 荈⹛䗍ⴓ禸ך涪㾜ח״׶فؚٗٓيה׃ג剅ֽ׸ל⡦ד׮堣唒㷕统〳腉ח ResNet50: 26 million params ResNet101: 45 million params EfficientNet-B7: 66 million params VGG19: 144 million params 12-layer, 12-heads BERT: 110 million params 24-layer, 16-heads BERT: 336 million params GPT-2 XL: 1558 million params GPT-3: 175 billion params 植➿ך堣唒㷕统כ ⭙⦐ךػًٓة 荈歋䏝 ׾䭯אٌرٕ׾侧⼧♰ך如⯋׾䭯א 侧⼪♰⦐ךر٦ةחؿ؍حذ؍ؚٝ׃גְג湫䠬ָ⸬ַזְꬊ荈僇ז朐屣

Slide 46

Slide 46 text

/ 166 ➙傈ךذ٦و 34 ˖ 荈䊹稱➜ 堣唒㷕统ה荈搫猰㷕ך㞮歲 ˖ 堣唒㷕统הכ倜׃ְفؚٗٓىؚٝך倯岀 ˖ 堣唒㷕统㾊כ♧⡤⡦ָ嚂׃ְךַ ˖ ⴓ㶨ך邌植ה堣唒㷕统 ˖ ؚٖ؎نحؙأ剑黝⻉ 怴糊䌓秛 锷椚㷕ה窟鎘㷕ך輐さ ˖ 荈搫猰㷕灇瑔ד堣唒㷕统׾⢪ֲֶהׅ׷ה䗳׆עאַ׷劤䔲חꨇ׃ְ㉏겗 ˖ ر٦ةٌرؚٔٝה✮庠،ٕ؞ٔؤي 5IF5XP$VMUVSFT ˖ ✮庠ַ椚鍑ַ3BTIPNPO⸬卓 6OEFSTQFDJDBUJPO 鍑ꅸ㢳圫䚍 ˖ ➂꟦ך钠濼غ؎،أח歋勻ׅ׷㉏겗⟎铡ծ㣟侁ծ䧭⸆غ؎،أծFUD ˖ 堣唒㷕统ַ׵堣唒涪鋅פ ˖ չ涪鋅պכさ椚⻉דֹ׷ךַׁ׵ח荈⹛⻉דֹ׷ךַ

Slide 47

Slide 47 text

/ 166 ⴓ㶨כչ穈さׇ涸պז⩎꬗׾׮א 35 https://cen.acs.org/physical-chemistry/computational-chemistry/Exploring-chemical-space-AI-take/98/i13

Slide 48

Slide 48 text

/ 166 ⴓ㶨鎸鶢㶨 36 NPSFUIBO EFTDSJQUPST ˖ %鎸鶢㶨 ˖ DPOTUJUVUJPOBMEFTDSJQUPST ˖ DPVOUEFTDSJQUPST ˖ %鎸鶢㶨 ˖ MJTUPGTUSVDUVSBMGSBHNFOUT ˖ GJOHFSQSJOUT ˖ %鎸鶢㶨 ˖ HSBQIJOWBSJBOUT ˖ %鎸鶢㶨 ˖ %.P34& 8)*. (&5"8": ˖ RVBOUVNDIFNJDBMEFTDSJQUPST ˖ TJ[F TUFSJD TVSGBDF WPMVNF FUD ˖ %鎸鶢㶨 ˖ (3*% $P.'" 7PMTVSG %3"(0/ EFTDSJQUPST ㉀欽ך鎸鶢㶨اؿزؐؑ، ˖ 㹋꿀涸ז鎘庠ꆀ ˖ 鎘皾涸ז鎸鶢㶨 Quantitative Structure–Property Relationship Modeling of Diverse Materials Properties. Chem Rev, 2012, 112 (5), pp 2889–2919 ؔ٦فٝا٦أؿٖ٦يٙ٦ؙ • Descriptors • Descriptors3D • GraphDescriptors • Fingerprints • ChemicalFeatures • ChemicalForceFields rdkit.Chem rdkit.ML.Descriptors Todeschini and Consonni, Molecular Descriptors for Chemoinformatics. Wiley‐VCH, 2009. https://doi.org/10.1002/9783527628766

Slide 49

Slide 49 text

/ 166 ⴓ㶨ךչ葺ְ害欽涸邌植պך㷕统 37 Reactions Materials Molecules CC1CCNO1 NCc1ccoc1.S=(Cl)Cl>>[RX_5]S=C=NCc1ccoc1 悵㖈㢌侧邌植 زهٗآ 갥挿暴䗙 鴟暴䗙 Representation Learning … ˖ ⴓ겲 ˖ 㔐䌓 ˖ 欰䧭 圫ղז ♴崧ةأؙ ⴓ㶨ך橆㞮勴⟝垥涸湱✼⡲欽瘝ך䞔㜠 ؚٓؿ邌植 Task-Specific Head

Slide 50

Slide 50 text

/ 166 6TF$BTF7JSUVBM4DSFFOJOH 24"32413 38 • Mutagenic potency • Carcinogenic potency • Endocrine disruption • Growth inhibition • Aqueous solubility N NH O O H H H H H H H H H H H H H H H H H H H H H H H H H O O O O O O Cl H H H H H H H H H H H H H H H H H Br Br O P O O Br Br O Br Br H H H H H H H H H H H H H H H N S N N H H H H H H H H H H H H H H H O N O O H H H O O H H N O O Cl Cl Cl H H H H H H H N O O H H H H H H H H H N O O H H H H H H H N H N O O N O O H H H H H H H H N CH3 O O H N Cl Cl Cl Cl Cl H 3 C O O O O O O H 3 C CH3 CH2 O HN O O NH CH3 HO OH CH 3 N O O CH 3 N N H N H H 3 C N H 3 C H 3 C NH O N O N O CH3 O N NH 2 O CH3 Br CH3 N H 3 C H N S N O CH3 N OH CH3 CH3 N N N CH3 H 3 C H2 N NH2 H OH O HO CH 3 H H O CH 3 H O O H 3 C H H H O H 3 C S CH3 O H H O CH3 CH3 O O HO H 3 C H HO F H O H 3 C NH 2 O N HO H O O H H O O O H 3 C O O O CH 3 O CH 3 H O CH 3 H O O CH 3 H H N H N O H 3 C O O O

Slide 51

Slide 51 text

/ 166 6TF$BTF7JSUVBM4DSFFOJOH 24"32413 39 https://pubchem.ncbi.nlm.nih.gov/bioassay/1

Slide 52

Slide 52 text

/ 166 6TF$BTF7JSUVBM4DSFFOJOH 24"32413 40 input output ML activity: “Active” LogGI50: -7.8811 CID 11978790 GI50: concentration required for 50% inhibition of growth

Slide 53

Slide 53 text

/ 166 .PMFDVMBS(SBQITⴓ㶨ךؚٓؿ邌植 41 Input representation (molecular graph) 1 2 1 3 explicit Hs 4 5 6 7 8 9 10 11 12 13 14 15 16 17 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Any permutation of this numbering should not change the results. ❗ CID 204 atoms → nodes bonds → edges 1. permutation equivariance 2. permutation invariance

Slide 54

Slide 54 text

/ 166 .PMFDVMBS(SBQITⴓ㶨ךؚٓؿ邌植 41 Input representation (molecular graph) 1 2 1 3 explicit Hs 4 5 6 7 8 9 10 11 12 13 14 15 16 17 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Any permutation of this numbering should not change the results. ❗ CID 204 • atomic_num (one-hot, 101) • total_degree (one-hot, 7) • formal_charge (one-hot, 6) • chiral_tag (one-hot, 5) • num_Hs (one-hot, 6) • hybridization (one-hot, 6) • is_aromatic (binary, 1) • atomic_mass (real, 1) 17 edge(bond) features • no_bond (binary, 1) • is_single (binary, 1) • is_double (binary, 1) • is_triple (binary, 1) • is_aromatic (binary, 1) • is_connjugated (binary, 1) • is_in_ring (binary, 1) • stereo (one-hot, 7) 17 14 133 node(atom) features 133 features 14 features e.g. Features for ChemProp (Yang et al, 2019) atoms → nodes bonds → edges 1. permutation equivariance 2. permutation invariance

Slide 55

Slide 55 text

/ 166 .PMFDVMBS(SBQITⴓ㶨ךؚٓؿ邌植 41 Input representation (molecular graph) 1 2 1 3 explicit Hs 4 5 6 7 8 9 10 11 12 13 14 15 16 17 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Any permutation of this numbering should not change the results. ❗ atom features bond features topology Molecular Graph read out graph-level output • sum, mean or max • attentive pooling CID 204 • atomic_num (one-hot, 101) • total_degree (one-hot, 7) • formal_charge (one-hot, 6) • chiral_tag (one-hot, 5) • num_Hs (one-hot, 6) • hybridization (one-hot, 6) • is_aromatic (binary, 1) • atomic_mass (real, 1) 17 edge(bond) features • no_bond (binary, 1) • is_single (binary, 1) • is_double (binary, 1) • is_triple (binary, 1) • is_aromatic (binary, 1) • is_connjugated (binary, 1) • is_in_ring (binary, 1) • stereo (one-hot, 7) 17 14 133 node(atom) features 133 features 14 features e.g. Features for ChemProp (Yang et al, 2019) atoms → nodes bonds → edges 1. permutation equivariance 2. permutation invariance

Slide 56

Slide 56 text

/ 166 (SBQI/FVSBM/FUXPSLT (//T 42 N O C C C C H H H H H N O C C C C H H H H H GNN Layer GNN updates features

Slide 57

Slide 57 text

/ 166 (SBQI/FVSBM/FUXPSLT (//T 42 N O C C C C H H H H H N O C C C C H H H H H GNN Layer GNN updates features

Slide 58

Slide 58 text

/ 166 (SBQI/FVSBM/FUXPSLT (//T 42 N O C C C C H H H H H N O C C C C H H H H H GNN Layer GNN updates features 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 hi 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 eij atom features bond features

Slide 59

Slide 59 text

/ 166 (SBQI/FVSBM/FUXPSLT (//T 42 N O C C C C H H H H H N O C C C C H H H H H GNN Layer GNN updates features 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 hi 0 @hi, M j2Ni (hi, hj, eij) 1 A Update by “Message Passing” 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 hi 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 eij atom features bond features

Slide 60

Slide 60 text

/ 166 (SBQI/FVSBM/FUXPSLT (//T 42 N O C C C C H H H H H N O C C C C H H H H H GNN Layer GNN updates features Ԯ ԮMessage Permutation equivariant operations • nn.Linear Bond features can be used (typically in Ԯ) AAAC/HichVHNTttAEB4b2tI0lNBeKnGxiKiChKJNhVrUU1QuPaHwE0DCkbU2m3jD+kfrTVBquQ/AC3DgVNQeEOIKD9BLX4ADBx4AcUylXnroxHEVtah0LHu//Wa+8bc7dih4pAi50vSx8QcPH008zj3JTz6dKkw/24yCjnRY3QlEILdtGjHBfVZXXAm2HUpGPVuwLXtveZDf6jIZ8cDfUL2QNTza8nmTO1QhZRX2TduL3cTihilYU1Epg33DDKNsX/qdXsiZNm8FoehEVtw2TO4bpkeV61ARr2A+QZHLjVG9kaH2ELHEink7mc+ZkrdcNW8ViqRM0jDugkoGipBFLShcgAm7EIADHfCAgQ8KsQAKET47UAECIXINiJGTiHiaZ5BADrUdrGJYQZHdw28LdzsZ6+N+0DNK1Q7+ReArUWnAHLkkJ6RPvpFTckN+/rNXnPYYeOnhag+1LLSmDl6s//ivysNVgTtS3etZQROWUq8cvYcpMziFM9R3Pxz219+uzcUvyTG5Rf+fyBX5iifwu9+dL6ts7egePzZ6wRvDAVX+HsddsPmqXHldXlxdLFbfZaOagBmYhRLO4w1U4T3UoI79r7UxLa9N6h/1z/qpfjYs1bVM8xz+CP38F+WDvZo= hi 0 @hi, M j2Ni (hi, hj, eij) 1 A Update by “Message Passing” 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 hi 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 eij atom features bond features

Slide 61

Slide 61 text

/ 166 (SBQI/FVSBM/FUXPSLT (//T 42 N O C C C C H H H H H N O C C C C H H H H H GNN Layer GNN updates features • sum, mean or max • attentive pooling ԯ ԯAggregate Permutation invariant operations Ԯ ԮMessage Permutation equivariant operations • nn.Linear Bond features can be used (typically in Ԯ) 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 hi 0 @hi, M j2Ni (hi, hj, eij) 1 A Update by “Message Passing” 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 hi AAACjnichVHLSsNAFL2Nr1ofjboR3BRLxVWZSKkiiEU3XfZhH9CWksSpjs2LJC3U0B9wLy4ERcGF+AF+gBt/wEU/QVxWcOPCmzQgWqw3TObMmXvunJkrGQqzbEJ6AW5sfGJyKjgdmpmdmw/zC4tFS2+ZMi3IuqKbZUm0qMI0WrCZrdCyYVJRlRRakpr77n6pTU2L6dqB3TFoTRWPNNZgsmgjValKqkO7dYeddOt8lMSJF5FhIPggCn5kdP4RqnAIOsjQAhUoaGAjVkAEC78KCEDAQK4GDnImIubtU+hCCLUtzKKYISLbxP8Rrio+q+HarWl5ahlPUXCYqIxAjLyQe9Inz+SBvJLPP2s5Xg3XSwdnaaClRj18tpz/+Fel4mzD8bdqpGcbGrDleWXo3fAY9xbyQN8+vejnt3MxZ43ckjf0f0N65AlvoLXf5bsszV2O8COhF3wxbJDwux3DoLgRF5LxRDYRTe35rQrCCqzCOvZjE1KQhgwUvBc9hyu45nguye1wu4NULuBrluBHcOkvYEmUlg== eij atom features bond features

Slide 62

Slide 62 text

/ 166 (SBQI/FVSBM/FUXPSLT (//T 42 N O C C C C H H H H H N O C C C C H H H H H GNN Layer GNN updates features • sum, mean or max • attentive pooling ԯ ԯAggregate Permutation invariant operations ԰ ԰Update Any • nn.Linear Ԯ ԮMessage Permutation equivariant operations • nn.Linear Bond features can be used (typically in Ԯ) AAAC/HichVHNTttAEB4b2tI0lNBeKnGxiKiChKJNhVrUU1QuPaHwE0DCkbU2m3jD+kfrTVBquQ/AC3DgVNQeEOIKD9BLX4ADBx4AcUylXnroxHEVtah0LHu//Wa+8bc7dih4pAi50vSx8QcPH008zj3JTz6dKkw/24yCjnRY3QlEILdtGjHBfVZXXAm2HUpGPVuwLXtveZDf6jIZ8cDfUL2QNTza8nmTO1QhZRX2TduL3cTihilYU1Epg33DDKNsX/qdXsiZNm8FoehEVtw2TO4bpkeV61ARr2A+QZHLjVG9kaH2ELHEink7mc+ZkrdcNW8ViqRM0jDugkoGipBFLShcgAm7EIADHfCAgQ8KsQAKET47UAECIXINiJGTiHiaZ5BADrUdrGJYQZHdw28LdzsZ6+N+0DNK1Q7+ReArUWnAHLkkJ6RPvpFTckN+/rNXnPYYeOnhag+1LLSmDl6s//ivysNVgTtS3etZQROWUq8cvYcpMziFM9R3Pxz219+uzcUvyTG5Rf+fyBX5iifwu9+dL6ts7egePzZ6wRvDAVX+HsddsPmqXHldXlxdLFbfZaOagBmYhRLO4w1U4T3UoI79r7UxLa9N6h/1z/qpfjYs1bVM8xz+CP38F+WDvZo= hi 0 @hi, M j2Ni (hi, hj, eij) 1 A Update by “Message Passing” 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 hi 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 eij atom features bond features

Slide 63

Slide 63 text

/ 166 6TF$BTF7JSUVBM4DSFFOJOH 24"32413 43 ChemProp (Directed MPNN) ExtraTrees w/ ECFP6(1024) Performance for unseen (test) data: Standard ML GNN Stokes et al, Cell (2020) https://doi.org/10.1016/j.cell.2020.01.021 Marchant, Nature (2020) https://doi.org/10.1038/d41586-020-00018-3 ChemProp (Yang et al, 2019) from MIT MLPDS (Machine Learning for Pharmaceutical Discovery and Synthesis) Consortium Disclaimer: This is just for a toy demo. This should be taken as classification for ACTIVITY_OUTCOME (Active or Inactive) 95.079% (Active/Inactive) 95.604% (Active/Inactive) • Regression for LogGI50 • Regression for LogGI50 • Classification accuracy • Classification accuracy RMSE 0.6076 RMSE 0.7970 Activie/Inactive (Classification), LogGI50 (Regression)

Slide 64

Slide 64 text

/ 166 &$'1ה/FVSBM(SBQI'JOHFSQSJOU 44 ˖ /FVSBM(SBQI'JOHFSQSJOU剑ⴱ劍ח䲿周ׁ׸׋(//ך♧א ˖ (SBQI$POWPMVUJPO׾欽ְ׋(//ך♧珏ה׫זׇ׷ ˖ &$'1 $JSDVMBS'JOHFSQSJOU ך'JOHFSQSJOU鎘皾׾ػًٓة׾䭯א䗍ⴓ〳腉ז 怴皾ד剅ֹ湫ֿׅהד䖤׵׸׷㷕统〳腉ז'JOHFSQSJOUהְֲ⡘縧בֽ Duvenaud, Maclaurin, Aguilera-Iparraguirre, Gómez-Bombarell, Hirzel, Aspuru-Guzik, Adams, Convolutional networks on graphs for learning molecular fingerprints. NIPS (2015)

Slide 65

Slide 65 text

/ 166 ("5ה5SBOTGPSNFS㘗(// 45 (Multihead) Self-attention Feed-forward NN Add + LayerNorm Add + LayerNorm ˖ ぐ갥挿ך暴䗙كؙزٕ׾刿倜ׅ׷ꥷח"UUFOUJPO׾Ⰵ׸׋ְ ˖ 5SBOTGPSNFSכزهٗآⵖ秈ךזְ(SBQI"UUFOUJPO/FUXPSL ("5 㢌珏ה׫זׇ׷ ˖ 鷞ח׮׍׹׿5SBOTGPSNFS㘗ך4FMG"UUFOUJPO׾(//ח׮׍ֿ׬ֿה׮דֹ׷ Transformer GNN Layer ⡂גְ׷˘ Embedding + Pos Encoding A Generalization of Transformer Networks to Graphs Dwivedi & Bresson (2020) https://arxiv.org/abs/2012.09699 Do Transformers Really Perform Bad for Graph Representation? Ying et al (2021) https://arxiv.org/abs/2106.05234 Communicative Representation Learning on Attributed Molecular Graphs Song et al (2020) https://www.ijcai.org/proceedings/2020/0392.pdf Graph-BERT: Only Attention is Needed for Learning Graph Representations Zhang et al (2020) https://arxiv.org/abs/2001.05140 Veličković, Cucurull, Casanova, Romero, Liò, Bengio, Graph Attention Networks (ICLR 2018) https://arxiv.org/abs/1710.10903 Joshi, Transformers are Graph Neural Networks. (2020) https://graphdeeplearning.github.io/post/transformers-are-gnns/ Ying et al (2021) ͷGraphormer͸ KDDCup 2021ͷOpen Graph Benchmark Large-Scale Challenge(ޙड़)ͷGraph-level λεΫͷ༏উϞσϧͰ࢖ΘΕͨ େن໛σʔλͳΒάϥϑͰ΋ Transformer͸༗ޮ…!?

Slide 66

Slide 66 text

/ 166 ⴓ㶨邌植ך✲⵸㷕统ה鯄獳㷕统 46 ˖ 5SBOTGPSNFSפךꟼ䗰כ 4FMG4VQFSWJTFEז 㣐鋉垷✲⵸㷕统ה鯄獳פך劍䖉ך植׸ ˖ ⴓ㶨ةأؙ׮植㹋ך⦐ⴽ朐屣דכ㼭؟ٝفٕד֮׷ֿהָקה׿ו ˖ ׮׃害欽ךⴓ㶨邌植׾㣐鋉垷✲⵸㷕统ח״׶栻䖤׃'FXTIPU;FSPTIPU鯄獳ָדֹ׷ ךז׵岚⿹⸬卓כ鎘׶濼׸זְ DG$7ך*NBHF/FUQSFUSBJOFE$// /-1ך#&35瘝 Strategies for Pre-training Graph Neural Networks Hu, Liu, Gomes, Zitnik, Liang, Pande, Leskovec (ICLR 2020) https://arxiv.org/abs/1905.12265 Self-Supervised Graph Transformer on Large-Scale Molecular Data Rong, Bian, Xu, Xie, Wei, Huang, Huang (NeurIPS 2020) https://arxiv.org/abs/2007.02835

Slide 67

Slide 67 text

/ 166 ⴓ㶨邌植ך欰䧭 47 ˖ ׮ֲמהאךⴓ㶨ך邌植㷕统פך劍䖉כⴓ㶨ؚٓؿװⴓ㶨圓鸡ך欰䧭 ˖ ⴓ㶨欰䧭ך㜥さכ暴ח%FDPEFSָꬊ荈僇ד圓鸡涸זⳢ椚׾㹋植ׅ׷䗳銲ָ֮׷ ˖ 圓䧭䚍ٌآُ٦ٕ䚍װ⻉㷕涸ٕ٦ٕ׮罋䣁׃זְה䠐㄂ךזְ⳿⸂חז׶䖤׷ ˖ 俑㶵⴨邌植 4.*-&4鎸岀 ַ׵ך欰䧭כ湫䱸涸זךדؚٓؿ邌植ך⮚⡘䚍׮銲嗚鏾 https://arxiv.org/abs/2012.15544

Slide 68

Slide 68 text

/ 166 6TF$BTF2VBOUVNDIFNJTUSZ 48 https://qcarchive.molssi.org/apps/ml_datasets/

Slide 69

Slide 69 text

/ 166 6TF$BTF2VBOUVNDIFNJTUSZ 49 input output gdb_21014 1000 sec Density Functional Theory (DFT) B3LYP/6-31G(2df, p) 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 ˆ H = E ⢽ ♧ꨵ㶨晛ך4DISµEJOHFS倯玎䒭 ,PIOˊ4IBN倯玎䒭 ך実鍑 QMܭࢉ

Slide 70

Slide 70 text

/ 166 6TF$BTF2VBOUVNDIFNJTUSZ 49 input output gdb_21014 1000 sec Density Functional Theory (DFT) B3LYP/6-31G(2df, p) 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 ˆ H = E ⢽ ♧ꨵ㶨晛ך4DISµEJOHFS倯玎䒭 ,PIOˊ4IBN倯玎䒭 ך実鍑 ML 0.01 sec ≈ 100,000 times faster! QMܭࢉ

Slide 71

Slide 71 text

/ 166 6TF$BTF2VBOUVNDIFNJTUSZ 50 ICML 2017 https://arxiv.org/abs/1704.01212 JCTC 2017 https://doi.org/10.1021/acs.jctc.7b00577 ˖ (PPHMFָ葿ղז(//ךغٔؒ٦ءّٝ׾չ.1// .FTTBHF1BTTJOH// պה׃ג 窟♧涸ח鋅湫׃׋ꥷחة٦؜حزחׁ׸׋ךָֿךꆀ㶨⻉㷕鎘皾鵚⡂ةأؙ

Slide 72

Slide 72 text

/ 166 6TF$BTF2VBOUVNDIFNJTUSZ 51 ؔٔآشٕך'$'1⾱㶨♶㢌ꆀ ؔٔآشٕך&$'1⾱㶨♶㢌ꆀ • the number of immediate neighbors who are “heavy” (non-hydrogen) atoms • the valence minus the number of hydrogens • the atomic number • the atomic mass • the atomic charge • the number of attached hydrogens • whether the atom is contained in at least one ring %BZMJHIU ⾱㶨♶㢌ꆀ • hydrogen-bond acceptor or not? • hydrogen-bond donor or not? • negatively ionizable or not? • positively ionizable or not? • aromatic or not? • halogen or not? Rogers and Hahn, JCIM (2005) https://doi.org/10.1021/ci100050t Faber et al, JCTC (2017) https://doi.org/10.1021/acs.jctc.7b00577 .1//ח״׷ꆀ㶨⻉㷕鎘皾鵚⡂ד欽ְ׵׸׋갥挿٥鴟暴䗙 鸬竲ꆀٓكٕ

Slide 73

Slide 73 text

/ 166 4DI/FU 52 input molecule H2O gdb_3 0 1 2 graph (SchNet) 0 1 atom features 0 1 2 2 edges w/ cutoff (10Å) 0 bond features 0 1 1 0 2 2 1 2 edge_index 0.9620 0.9622 1.5133 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 rij := kri rj k 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 Z0 = 8 AAACinichVHLSsNAFL2Nr1qrrboR3ARLxVWZlOIToagLl33YB1YpSZzWoWkSkmmhFn/AnSvBrhRciB/gB7jxB1z0E8RlBTcuvEkDosV6w2TOnLnnzpm5iqkxmxPS9Qkjo2PjE/7JwFRweiYUnp3L20bDUmlONTTDKiqyTTWm0xxnXKNF06JyXdFoQantOvuFJrVsZugHvGXS47pc1VmFqTJHqnBYlsRtUSqHIyRG3BAHgeSBCHiRMsKPcAQnYIAKDagDBR04Yg1ksPErgQQETOSOoY2chYi5+xTOIYDaBmZRzJCRreG/iquSx+q4dmrarlrFUzQcFipFiJIXck965Jk8kFfy+WettlvD8dLCWelrqVkOXSxkP/5V1XHmcPqtGuqZQwXWXa8MvZsu49xC7eubZ1e97GYm2l4mt+QN/d+QLnnCG+jNd/UuTTOdIX4U9IIvhg2SfrdjEOTjMWk1lkgnIskdr1V+WIQlWMF+rEES9iEFObf+JVxDRwgKcWFD2OqnCj5PMw8/Qtj7AiFSkVQ= Z1 = 1 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 Z2 = 1 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 r0 = [ 0.0344, 0.9775, 0.0076] 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 r1 = [0.0648, 0.0206, 0.0015] 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 r2 = [0.8718, 1.3008, 0.0007] SchNet (Schütt et al, 2017) AAAC9XichVHNShxBEK6d+BeNuiaXgJchi7IiLL0iieQk5pKT+Lcq2DL0jL27vfZMDz29a3SYF8gL5BByiKgQcvDoA3jxBRLwEYJHBS8erJ0dEqKoNcz011/VV/N1lxtKERlCznPWs67unt6+5/0DLwaHhvMjL1cj1dQer3hKKr3usohLEfCKEUby9VBz5ruSr7nbH9r5tRbXkVDBitkN+abPaoGoCo8ZpJx8SF0//pQ4wqaSVw3TWu3Yf7lJm4aRSDNFGjV9J27YVAQ29Zmpe0zG81iVYFFdFDNRY8KmakuZtAlVPq+xxIlFI6Fa1OpmwskXSImkYd8H5QwUIIsFlT8BClugwIMm+MAhAINYAoMInw0oA4EQuU2IkdOIRJrnkEA/aptYxbGCIbuN3xruNjI2wH27Z5SqPfyLxFej0oYx8ov8IJfkjPwkf8jNg73itEfbyy6ubkfLQ2f48+vl6ydVPq4G6v9Uj3o2UIWZ1KtA72HKtE/hdfStvS+Xy++XxuJxsk8u0P93ck5O8QRB68o7XORLXx/x46IXvDEcUPnuOO6D1alS+W1penG6MDuXjaoPRuENFHEe72AWPsICVLD/b7jJded6rB3rm3VgHXVKrVymeQX/hXV8C560vMU= xi xi + 0 @ X j2Ni (xj) !ij 1 A Message Passing with residual connections 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 x0 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 x1 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 x2 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 w01 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 w02 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 w12

Slide 74

Slide 74 text

/ 166 4DI/FU 52 input molecule H2O gdb_3 0 1 2 graph (SchNet) 0 1 atom features 0 1 2 2 edges w/ cutoff (10Å) 0 bond features 0 1 1 0 2 2 1 2 edge_index 0.9620 0.9622 1.5133 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 rij := kri rj k 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 Z0 = 8 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 Z1 = 1 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 Z2 = 1 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 r0 = [ 0.0344, 0.9775, 0.0076] AAACpnichVHLSsNQED3GV3226kZwUy2KCykTqbUIgujGlfiqFWqpSbxqMC+StKDFteAPuHCl4ELErX6AG3/AhZ8gLhXcuHCSBkRFnZDMuefOmZx7R3UM3fOJHhukxqbmltZYW3tHZ1d3PNHTu+bZFVcTec02bHddVTxh6JbI+7pviHXHFYqpGqKg7s0F+4WqcD3dtlb9fUeUTGXH0rd1TfGZKicGN1Sz5h6W5eR0skhpymZyY5zGKRskkidK5UQqQEEkfwI5AilEsWgnbrGBLdjQUIEJAQs+YwMKPH6KkEFwmCuhxpzLSA/3BQ7RztoKVwmuUJjd4+8Or4oRa/E66OmFao3/YvDrsjKJYXqgS3qhe7qiJ3r/tVct7BF42ees1rXCKceP+1fe/lWZnH3sfqr+9OxjG7nQq87enZAJTqHV9dWDk5eVqeXh2gid0zP7P6NHuuMTWNVX7WJJLJ/+4UdlL3xjPCD5+zh+grXxtJxNZ5YyqZnZaFQxDGAIozyPScxgHovIc/8jXOMGt9KotCDlpUK9VGqINH34EtLmBwvLmR0= r1 = [0.0648, 0.0206, 0.0015] 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 r2 = [0.8718, 1.3008, 0.0007] nn.Embedding 128 -1.249 1.6278 -0.1370 ⋮ -0.9488 0.3105 -1.6185 0.1960 ⋮ -0.5310 0.3105 -1.6185 0.1960 ⋮ -0.5310 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 x0 AAACi3ichVHLSsNAFL2Nr1qtrboR3BRLxVWZaFEpLooiuOzDPqAtJYnTOjQvkrRYQ3/ApRsXdaPgQvwAP8CNP+CinyAuK7hx4U0aEC3WGyZz5sw9d87MFXWZmRYhfR83MTk1PeOfDczNBxdC4cWlgqm1DInmJU3WjJIomFRmKs1bzJJpSTeooIgyLYrNA2e/2KaGyTT12OrotKoIDZXVmSRYSJUqomKfdWt8LRwlceJGZBTwHoiCF2kt/AgVOAENJGiBAhRUsBDLIICJXxl4IKAjVwUbOQMRc/cpdCGA2hZmUcwQkG3iv4GrssequHZqmq5awlNkHAYqIxAjL+SeDMgzeSCv5PPPWrZbw/HSwVkcaqleC12s5D7+VSk4W3D6rRrr2YI67LpeGXrXXca5hTTUt8+vBrlkNmavk1vyhv5vSJ884Q3U9rt0l6HZ3hg/InrBF8MG8b/bMQoKm3F+O57IJKKpfa9VfliFNdjAfuxACo4gDXm3D5fQg2suyG1xSW5vmMr5PM0y/Aju8AulVJLx x1 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 x2 SchNet (Schütt et al, 2017) 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 xi xi + 0 @ X j2Ni (xj) !ij 1 A Message Passing with residual connections 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 w01 AAACjnichVFLSwJRFD5OL7OHVpugjSRGKzmKWASR1Malj3yAisxMVxucFzOjYYN/oH20CIqCFtEP6Ae06Q+08CdES4M2LTqOA1GSneHO/e53z3fud+8RdFkyLcSeh5uYnJqe8c765uYXFv2BpeWCqbUMkeVFTdaMksCbTJZUlrckS2Yl3WC8IsisKDQPBvvFNjNMSVMPrY7OqgrfUKW6JPIWUeWKoNgn3ZqNsW4tEMIIOhEcBVEXhMCNtBZ4hAocgQYitEABBipYhGXgwaSvDFFA0Imrgk2cQUhy9hl0wUfaFmUxyuCJbdK/Qauyy6q0HtQ0HbVIp8g0DFIGIYwveI99fMYHfMXPP2vZTo2Blw7NwlDL9Jr/bDX38a9KodmC42/VWM8W1GHb8SqRd91hBrcQh/r26UU/t5MN2xt4i2/k/wZ7+EQ3UNvv4l2GZS/H+BHIC70YNSj6ux2joBCLRBOReCYeSu67rfLCGqzDJvVjC5KQgjTknRc9hyu45gJcgtvl9oapnMfVrMCP4FJflhyUNw== w02 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 w12

Slide 75

Slide 75 text

/ 166 4DI/FU 52 input molecule H2O gdb_3 0 1 2 graph (SchNet) 0 1 atom features 0 1 2 2 edges w/ cutoff (10Å) 0 bond features 0 1 1 0 2 2 1 2 edge_index 0.9620 0.9622 1.5133 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 rij := kri rj k 50 MLP 50 →128 0.1803 0.0826 0.3349 ⋮ -0.474 0.0403 -0.003 0.0802 ⋮ -0.061 0.1803 0.0826 0.3349 ⋮ -0.474 128 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 exp( ↵(rij µ↵)2) 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 Z0 = 8 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 Z1 = 1 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 Z2 = 1 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 r0 = [ 0.0344, 0.9775, 0.0076] AAACpnichVHLSsNQED3GV3226kZwUy2KCykTqbUIgujGlfiqFWqpSbxqMC+StKDFteAPuHCl4ELErX6AG3/AhZ8gLhXcuHCSBkRFnZDMuefOmZx7R3UM3fOJHhukxqbmltZYW3tHZ1d3PNHTu+bZFVcTec02bHddVTxh6JbI+7pviHXHFYqpGqKg7s0F+4WqcD3dtlb9fUeUTGXH0rd1TfGZKicGN1Sz5h6W5eR0skhpymZyY5zGKRskkidK5UQqQEEkfwI5AilEsWgnbrGBLdjQUIEJAQs+YwMKPH6KkEFwmCuhxpzLSA/3BQ7RztoKVwmuUJjd4+8Or4oRa/E66OmFao3/YvDrsjKJYXqgS3qhe7qiJ3r/tVct7BF42ees1rXCKceP+1fe/lWZnH3sfqr+9OxjG7nQq87enZAJTqHV9dWDk5eVqeXh2gid0zP7P6NHuuMTWNVX7WJJLJ/+4UdlL3xjPCD5+zh+grXxtJxNZ5YyqZnZaFQxDGAIozyPScxgHovIc/8jXOMGt9KotCDlpUK9VGqINH34EtLmBwvLmR0= r1 = [0.0648, 0.0206, 0.0015] AAACpnichVHLSsNAFD3G97NVN4KbalFcSLjRYosgiG5cia9aoZaaxFGDeZGkBS2uBX/AhSsFFyJu7Qe48Qdc+AniUsGNC2/TgKioN0zmzJl77pyZq7mm4QdEjw1SY1NzS2tbe0dnV3dPLN7bt+47JU8XWd0xHW9DU31hGrbIBkZgig3XE6qlmSKn7c/X9nNl4fmGY68FB64oWOqubewYuhowVYwPbWpWxTsqTiRmEnmSM2klM67Ik0SZcZKJKF0oxpMh4kj8BEoEkohiyYlXsYltONBRggUBGwFjEyp8/vJQQHCZK6DCnMfICPcFjtDB2hJnCc5Qmd3n/y6v8hFr87pW0w/VOp9i8vBYmcAIPdAVvdA9XdMTvf9aqxLWqHk54Fmra4VbjJ0MrL79q7J4DrD3qfrTc4AdZEKvBnt3Q6Z2C72uLx+evqxOr4xURumCntn/OT3SHd/ALr/ql8ti5ewPPxp74RfjBinf2/ETrE/IypScWk4lZ+eiVrVhEMMY436kMYsFLCHL9Y9xg1tUpTFpUcpKuXqq1BBp+vElpK0PJf+ZKQ== r2 = [0.8718, 1.3008, 0.0007] nn.Embedding 128 -1.249 1.6278 -0.1370 ⋮ -0.9488 0.3105 -1.6185 0.1960 ⋮ -0.5310 0.3105 -1.6185 0.1960 ⋮ -0.5310 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 x0 AAACi3ichVHLSsNAFL2Nr1qtrboR3BRLxVWZaFEpLooiuOzDPqAtJYnTOjQvkrRYQ3/ApRsXdaPgQvwAP8CNP+CinyAuK7hx4U0aEC3WGyZz5sw9d87MFXWZmRYhfR83MTk1PeOfDczNBxdC4cWlgqm1DInmJU3WjJIomFRmKs1bzJJpSTeooIgyLYrNA2e/2KaGyTT12OrotKoIDZXVmSRYSJUqomKfdWt8LRwlceJGZBTwHoiCF2kt/AgVOAENJGiBAhRUsBDLIICJXxl4IKAjVwUbOQMRc/cpdCGA2hZmUcwQkG3iv4GrssequHZqmq5awlNkHAYqIxAjL+SeDMgzeSCv5PPPWrZbw/HSwVkcaqleC12s5D7+VSk4W3D6rRrr2YI67LpeGXrXXca5hTTUt8+vBrlkNmavk1vyhv5vSJ884Q3U9rt0l6HZ3hg/InrBF8MG8b/bMQoKm3F+O57IJKKpfa9VfliFNdjAfuxACo4gDXm3D5fQg2suyG1xSW5vmMr5PM0y/Aju8AulVJLx x1 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 x2 Weighted ACSFs (ACSFs = atom- centered symmetry functions) for Behler-Parrinello potentials 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 1 2 ✓ cos ✓ ⇡rij rc ◆ + 1 ◆ cutoff function element-wise product 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 w01 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 w02 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 w12 SchNet (Schütt et al, 2017) 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 xi xi + 0 @ X j2Ni (xj) !ij 1 A Message Passing with residual connections

Slide 76

Slide 76 text

/ 166 6TF$BTF2VBOUVNDIFNJTUSZ 53 pred vs true for SchNet (Schütt et al, 2017) pred vs true for DimeNet (Klicpera et al, 2020) Dipole Moment Energy U HOMO LUMO Heat Capacity Enthalpy H Dipole Moment Energy U HOMO LUMO Heat Capacity Enthalpy H

Slide 77

Slide 77 text

/ 166 6TF$BTF2VBOUVNDIFNJTUSZ 53 pred vs true for SchNet (Schütt et al, 2017) pred vs true for DimeNet (Klicpera et al, 2020) Dipole Moment Energy U HOMO LUMO Heat Capacity Enthalpy H Dipole Moment Energy U HOMO LUMO Heat Capacity Enthalpy H ⦓鸞ְז׵ ⼧ⴓ鏩㺁דֹ׷✮庠铎䊴 ֿ׸כذأزر٦ة 鎮箺儗ח 鋅ׇגזְر٦ة ך穠卓

Slide 78

Slide 78 text

/ 166 6TF$BTF2VBOUVNDIFNJTUSZ 54 SchNet (Schütt et al, 2017) DimeNet (Klicpera et al, 2020) Free Energy Free Energy y_true y_true y_pred y_pred ExtraTrees w/ ECFP6 LightGBM w/ ECFP6 3-Layer MLP w/ ECFP6 (without 3D geometry) (without 3D geometry) (without 3D geometry) Free Energy Free Energy Free Energy

Slide 79

Slide 79 text

/ 166 4DI/0SC岚⹛ꟼ侧荈⡤׾堣唒㷕统 55

Slide 80

Slide 80 text

/ 166 (//TGPS(FPNFUSJD%FFQ-FBSOJOH 56 https://arxiv.org/abs/2104.13478 https://youtu.be/uF53xsT7mjc https://youtu.be/w6Pw4MOzMuo ICLR 2021 Keynote (Michael Bronstein) Seminar Talk (Petar Veličković) (//כ䌴䎢ְ䎗⡦圓鸡׾窟♧涸ח䪔ִ׷単穈׫ 堣唒㷕统ךٕؒٓٝ؜ٝ٥فؚٗٓي (T(SJET (SPVQT (SBQIT (FPEFTJDT(BVHFT

Slide 81

Slide 81 text

/ 166 ِ٦ؙٔحسך麊⹛纇חꟼׅ׷♶㢌䚍٥ず㢌䚍 57 ˖ ِ٦ؙٔحس纇& %ך⚛鹌٥㔐鯄㼎獥䚍 ˖ 暴婊ِ٦ؙٔحس纇4& %ך⚛鹌٥㔐鯄٥ꖎ⫷㼎獥䚍 Schütt et al, SchNet. (2017) https://arxiv.org/abs/1706.08566 Satorras et al, E(n) Equivariant Graph Neural Networks. (2021) https://arxiv.org/abs/2102.09844 Anderson et al, Cormorant. (2019) https://arxiv.org/abs/1906.04015 Unke et al, PhysNet. (2019) https://arxiv.org/abs/1902.08408 Klicpera et al, DimeNet++. (2020) https://arxiv.org/abs/2011.14115 Fuchs et al, SE(3)-Transformers. (2021) https://arxiv.org/abs/2006.10503 Köhler et al, Equivariant Flows (Radial Field). (2020) https://arxiv.org/abs/2006.02425 Thomas et al, Tensor Field Networks. (2018) https://arxiv.org/abs/1802.08219 ⱖ⫷ָ㢌䳔חꟼ׃ג ♶㢌 JOWBSJBU ず㢌 FRVJWBSJBOU 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 f(g · x) = g · f(x) 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 f(g · x) = f(x) 㢌䳔׃ג׮׃זְהֹה㢌׻׵זְ 㢌䳔׃גַ׵ⱖ⫷׃ג׮ⱖ⫷׃גַ׵㢌䳔׃ג׮㢌׻׵זְ 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 g 2 G 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 f : X ! Y 䎗⡦涸(//דכ㛇劤涸ז銲锜 暴חꆀ㶨⻉㷕鎘皾鵚⡂ך㜥さ & ず㢌 & ♶㢌 4& ず㢌 ⾱㶨ךYZ[䏟垥⦼׾׉ךתת갥挿暴䗙ꆀחׅ׷ךכ– ➿周ך⢽⾱㶨꟦騃ꨄ׾鴟暴䗙ח 䎂遤獳⹛װ㔐鯄דYZ[כ㢌׻׷ָ⢽ִל׉ךⴓ㶨ךؒطؘٕ٦כ㢌׻׵זְ 갥挿װ鴟ך暴䗙ꆀװ(// ⱖ⫷ ךرؠ؎ٝד㹋植ׅ׷

Slide 82

Slide 82 text

/ 166 ⴓ㶨邌植ך窟♧涸倯岀锷חז׶ִ׷ַ 58 ػة٦ٝ鎉铂ה׃ג ⻉㷕ך侄猰剅٥ر٦ةك٦أח֮׷濼陎邌植 Brc1cncc(Br)c1 C[O-] CN(C)C=O Na+ COc1cncc(Br)c1 SMILES Structural Formla Steric Structures Electronic States Reactants Reagents Products 暟椚涸㼎韋ה׃ג ꆀ㶨⻉㷕ח㛇בֻꨵ㶨朐䡾鎘皾 俑㶵⴨ ؚٓؿ %挿꧊さ ⴓ䋒PS 7PMVNF

Slide 83

Slide 83 text

/ 166 0(#-BSHF4DBMF$IBMMFOHF ,%%$VQ 59 1st place: 10 GNNs (12-Layer Graphormer) + 8 ExpC*s (5-Layer ExpandingConv) 73 GNNs (11-Layer LiteGEMConv with Self-Supervised Pretraining) 20 GNNs (32-Layer GNN with Noisy Nodes) Test MAE 0.1200 (eV) 2nd place:Test MAE 0.1204 (eV) 3rd place: Test MAE 0.1205 (eV) %ךⴓ㶨ؚٓؿַ׵ꆀ㶨⻉㷕鎘皾 %'5鎘皾 ד実׭׋)0.0-6.0ٍؘحف׾✮庠ׅ׷ةأؙ ر٦ةإحز1VC$IFN2$ַ׵ ؚٓؿ DG2.כ ؚٓؿ Results: https://ogb.stanford.edu/kddcup2021/results/#awardees_pcqm4m https://ogb.stanford.edu/kddcup2021/

Slide 84

Slide 84 text

/ 166 *$3F%%⻉㷕⿾䘔ךرؠ؎ٝה涪鋅 60 "'*3 (33. "SUJDJBM'PSDF*OEVDFE3FBDUJPO ➂䊨⸂钷饯⿾䘔岀 (MPCBM3FBDUJPO3PVUF.BQQJOH ؚٗ٦غٕ⿾䘔穗騟وحؾؚٝ ⚥呌䪮遭כ䬿挿ꞿ٥⵸歊椚侄䱇ָꟚ涪׃׋ꆀ㶨⻉㷕鎘皾ח㛇בֻ ⻉㷕⿾䘔穗騟荈⹛䱱稊،ٕ؞ٔؤي S. Maeda, Y. Harabuchi, Exploring paths of chemical transformations in molecular and periodic systems: An approach utilizing force., WIREs Comput. Mol. Sci., 2021, 11, e1538. https://doi.org/10.1002/wcms.1538 ㉀欽晛اؿزؐؑ،ךせ⵸ח׮

Slide 85

Slide 85 text

/ 166 䞔㜠猰㷕ַ׵鋅׋⿾䘔穗騟ך荈⹛䱱稊 61 㸜㹀朐䡾 &2 㸜㹀朐䡾 &2 鼂獳朐䡾 54 θ 1 θ 2 鍑ֻץֹ㉏겗 㢳䂼䚍ך㢳㢌侧ꟼ侧ך 噰㼭鍑הꭓ挿ךչⰋ⴨䮙պ ׉ך⾱㶨ꂁ縧Ăח֮׷הֹך Ⰻؒطؘٕ٦& Ă ך⦼

Slide 86

Slide 86 text

/ 166 ⵖ秈 ꟼ侧䕎כꤿחכ剅ֽ׆ծ ぐ挿ךꟼ侧⦼G Y כ 䭷㹀׃׋Y׀הח鎘皾儗꟦׾ 銲ׅ׷㉏ְさ׻ׇד䖤׷ 䞔㜠猰㷕ַ׵鋅׋⿾䘔穗騟ך荈⹛䱱稊 62 㸜㹀朐䡾 &2 㸜㹀朐䡾 &2 鼂獳朐䡾 54 θ 1 θ 2 鍑ֻץֹ㉏겗 㢳䂼䚍ך㢳㢌侧ꟼ侧ך 噰㼭鍑הꭓ挿ךչⰋ⴨䮙պ ׉ך⾱㶨ꂁ縧Ăח֮׷הֹך Ⰻؒطؘٕ٦& Ă ך⦼  

Slide 87

Slide 87 text

/ 166 䞔㜠猰㷕ַ׵鋅׋⿾䘔穗騟ך荈⹛䱱稊 63 㸜㹀朐䡾 &2 㸜㹀朐䡾 &2 鼂獳朐䡾 54 θ 1 θ 2 鍑ֻץֹ㉏겗 㢳䂼䚍ך㢳㢌侧ꟼ侧ך 噰㼭鍑הꭓ挿ךչⰋ⴨䮙պ ׉ך⾱㶨ꂁ縧Ăח֮׷הֹך Ⰻؒطؘٕ٦& Ă ך⦼   ⾱㶨ꂁ縧 HFPNFUSZ 1PUFOUJBM FOFSHZ ꨵ㶨朐䡾 岚⹛ꟼ侧 ꆀ㶨⻉㷕鎘皾 ˖ (BVTTJBO ˖ 03$" ˖ HBNFTT

Slide 88

Slide 88 text

/ 166 䞔㜠猰㷕ַ׵鋅׋⿾䘔穗騟ך荈⹛䱱稊 63 㸜㹀朐䡾 &2 㸜㹀朐䡾 &2 鼂獳朐䡾 54 θ 1 θ 2 鍑ֻץֹ㉏겗 㢳䂼䚍ך㢳㢌侧ꟼ侧ך 噰㼭鍑הꭓ挿ךչⰋ⴨䮙պ ׉ך⾱㶨ꂁ縧Ăח֮׷הֹך Ⰻؒطؘٕ٦& Ă ך⦼   ⾱㶨ꂁ縧 HFPNFUSZ 1PUFOUJBM FOFSHZ ꨵ㶨朐䡾 岚⹛ꟼ侧

Slide 89

Slide 89 text

/ 166 䞔㜠猰㷕ַ׵鋅׋⿾䘔穗騟ך荈⹛䱱稊 63 㸜㹀朐䡾 &2 㸜㹀朐䡾 &2 鼂獳朐䡾 54 θ 1 θ 2 鍑ֻץֹ㉏겗 㢳䂼䚍ך㢳㢌侧ꟼ侧ך 噰㼭鍑הꭓ挿ךչⰋ⴨䮙պ ׉ך⾱㶨ꂁ縧Ăח֮׷הֹך Ⰻؒطؘٕ٦& Ă ך⦼   ⾱㶨ꂁ縧 HFPNFUSZ 1PUFOUJBM FOFSHZ ꨵ㶨朐䡾 岚⹛ꟼ侧 ꆀ㶨⻉㷕鎘皾 ˖ (BVTTJBO ˖ 03$" ˖ HBNFTT

Slide 90

Slide 90 text

/ 166 䞔㜠猰㷕ַ׵鋅׋⿾䘔穗騟ך荈⹛䱱稊 64 㸜㹀朐䡾 &2 㸜㹀朐䡾 &2 鼂獳朐䡾 54 θ 1 θ 2 鍑ֻץֹ㉏겗 㢳䂼䚍ך㢳㢌侧ꟼ侧ך 噰㼭鍑הꭓ挿ךչⰋ⴨䮙պ ׉ך⾱㶨ꂁ縧Ăח֮׷הֹך Ⰻؒطؘٕ٦& Ă ך⦼ ׉׿זך搀椚דכ 䎃ծ剣せז鎘皾⻉㷕ך侄猰剅 ד'SBOL+FOTFOכչ㢌侧ך侧ָ זְ׃׾馄ִ׷ה⻉㷕⿾䘔ך鼂 獳朐䡾׾ׅץג実׭׷ֿהכ♶〳 腉ד֮׷պה鶢ץ׋կ

Slide 91

Slide 91 text

/ 166 䞔㜠猰㷕ַ׵鋅׋⿾䘔穗騟ך荈⹛䱱稊 64 㸜㹀朐䡾 &2 㸜㹀朐䡾 &2 鼂獳朐䡾 54 θ 1 θ 2 鍑ֻץֹ㉏겗 㢳䂼䚍ך㢳㢌侧ꟼ侧ך 噰㼭鍑הꭓ挿ךչⰋ⴨䮙պ ׉ך⾱㶨ꂁ縧Ăח֮׷הֹך Ⰻؒطؘٕ٦& Ă ך⦼ ׉׿זך搀椚דכ 䎃ծ剣せז鎘皾⻉㷕ך侄猰剅 ד'SBOL+FOTFOכչ㢌侧ך侧ָ זְ׃׾馄ִ׷ה⻉㷕⿾䘔ך鼂 獳朐䡾׾ׅץג実׭׷ֿהכ♶〳 腉ד֮׷պה鶢ץ׋կ

Slide 92

Slide 92 text

/ 166 䞔㜠猰㷕ַ׵鋅׋⿾䘔穗騟ך荈⹛䱱稊 65 㸜㹀朐䡾 &2 㸜㹀朐䡾 &2 鼂獳朐䡾 54 θ 1 θ 2 鍑ֻץֹ㉏겗 㢳䂼䚍ך㢳㢌侧ꟼ侧ך 噰㼭鍑הꭓ挿ךչⰋ⴨䮙պ ˖ 噰㼭鍑 &2 ֶ״ןꭓ挿 54 ׾ 実׭׷̔չ"'*3պ ˖ Ⰻ⴨䮙٥ⵖ秈➰ֹ穗騟䱱稊̔ չ(33.պ 'SBOL+FOTFOך侄猰剅כ䎃ח չ㣐ꅿ٥⵸歊׵ך倯岀ד〳腉חז׏׋պ ה剅ֹ䳔ִ׵׸׋

Slide 93

Slide 93 text

/ 166 噰㼭鍑 &2 ֶ״ןꭓ挿 54 ׾実׭׷̔չ"'*3պ 66 ׉׮׉׮չ1&4պך㢳䂼䚍ָז׈欰ׄ׷ַ׾罋ִג׫׷ אך⾱㶨"ה#ָ⿾䘔׃גⴓ㶨"#ח ז׷הֹؒطؘٕ٦ꥺ㠖ָ植׸׷ַ׵ " # S "# "# &2 54 騃ꨄח䘔ׄ׋؜ة׾כַׇ׷ה וַֿד⽃锃幾㼰ꟼ侧חז׶ ꨇ׃ְ)JMM$MJNCJOHָ♶銲ח 騃ꨄח䘔ׄ׋ؒطؘٕ٦ ⸂ ׾ ➂䊨涸ח⸇ִ׷ֿהח湱䔲 鵚⡂54

Slide 94

Slide 94 text

/ 166 噰㼭鍑 &2 ֶ״ןꭓ挿 54 ׾実׭׷̔չ"'*3պ 67 ぐ⾱㶨׾⾱㶨꟦ך騃ꨄָ㼭ֻׁז׷倯ぢפ➂䊨⸂'ד䬃ׅ PS䒷ֻ "SUJDJBM'PSDF*OEVDFE3FBDUJPO ➂䊨⸂钷饯⿾䘔岀 ⾱㶨꧊さ4ה4ך꟦ך"'*3ꟼ侧 騃ꨄ ˖ ꦄ䱸&2ה穗歋54ָ 䖤׵׸׷ ˖ '׾禸窟涸ח㢌ִג ꦄ䱸&2׾笨繎涸ח䱱稊 ꦄ䱸&2 穗歋ׅ׷54 Ⱏ剣穠さ⼱䖇 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 rij AAAChnichVG7TgJBFD2sL8QXamNiQyQYKzIYFGNFtLHkIY8ECdldB5y47G52FxIk/oCJrRRWmlgYP8APsPEHLPgEY4mJjYWXZROjRLyb2Tlz5p47Z+YqpiZsh7GuTxobn5ic8k8HZmbn5heCi0t522hYKs+phmZYRUW2uSZ0nnOEo/GiaXG5rmi8oJzu9/cLTW7ZwtAPnZbJy3W5pouqUGWHqGymIirBMIsyN0LDIOaBMLxIGcFHHOEYBlQ0UAeHDoewBhk2fSXEwGASV0abOIuQcPc5zhEgbYOyOGXIxJ7Sv0arksfqtO7XtF21SqdoNCxShhBhL+ye9dgze2Cv7PPPWm23Rt9Li2ZloOVmZeFiJfvxr6pOs4OTb9VIzw6q2HG9CvJuukz/FupA3zzr9LK7mUh7nd2yN/J/w7rsiW6gN9/VuzTPXI/wo5AXejFqUOx3O4ZBfjMa247G0/Fwcs9rlR+rWMMG9SOBJA6QQo7q13CJK3QkvxSVtqTEIFXyeZpl/Agp+QV6ppCu Ri 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 ػًٓة ⸂'ך䓼ׁ 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 ↵ = Ⰵ⸂&2 4 4 ֿך鵚⡂涸ז♳歲⦼⟃♴ ך⿾䘔ꥺ㠖ד鼂獳דֹ׷ ⿾䘔穗騟׾笨繎涸ח䱱稊

Slide 95

Slide 95 text

/ 166 噰㼭鍑 &2 ֶ״ןꭓ挿 54 ׾実׭׷̔չ"'*3պ 68 ⳿涪暟颵ה欵暟ָ ♷ִ׵׸׷✳畭挿穗騟 ׾実׭׷㜥さח⢪ֲ "#̔9 さ⢘㘗さ䧭 "#̔9: 穈剏㘗さ䧭 "̔9 殯䚍⻉ "̔9: ⴓ鍑٥⛣ꨄ .$"'*3 4$"'*3 %4"'*3 ⸂ ⸂ "%%' 4$"'*3הずׄ湡涸׌ָ "'*3ה殯ז׷䩛岀 㢳䧭ⴓך꟦ח⸂׾ַֽ׷ ⽃♧䧭ⴓךⰻך⾱㶨㔚ך ꟦ח⸂׾ַֽ׷ ぐ⾱㶨׾⾱㶨꟦ך騃ꨄָ㼭ֻׁז׷倯ぢפ➂䊨⸂'ד䬃ׅ PS䒷ֻ "SUJDJBM'PSDF*OEVDFE3FBDUJPO ➂䊨⸂钷饯⿾䘔岀

Slide 96

Slide 96 text

/ 166 Ⰻ⴨䮙٥ⵖ秈➰ֹ穗騟䱱稊̔չ(33.պ 69 ⳿涪朐䡾ַ׵鼂獳〳腉ז&2׾"'*3ד⴨䮙׃ծ倜׋ז&2ָ⳿זֻז׷תד ⱄ䌓涸חֿךⳢ椚׾黝欽 (MPCBM3FBDUJPO3PVUF.BQQJOH ؚٗ٦غٕ⿾䘔穗騟وحؾؚٝ EQ 㹋ꥷחכ⾱㶨玎䏝ך禸דזְהⰋ⴨䮙כ 侧ָ㢳ֺׅג搀椚זךד黝㹅䗳銲ח䘔ׄגⵖ秈 ̔(333.דכ鸞䏝㹀侧遤⴨　簭岀 3$.$ ָⵃ欽〳腉 EQ EQ EQ EQ EQ EQ EQ EQ EQ EQ EQ EQ EQ EQ EQ EQ EQ EQ

Slide 97

Slide 97 text

/ 166 *$3F%%⻉㷕⿾䘔ךرؠ؎ٝה涪鋅 70 EQ1 EQ2 Chemical Reaction How we can have this?

Slide 98

Slide 98 text

/ 166 *$3F%%⻉㷕⿾䘔ךرؠ؎ٝה涪鋅 70 EQ1 EQ2 Chemical Reaction

Slide 99

Slide 99 text

/ 166 *$3F%%⻉㷕⿾䘔ךرؠ؎ٝה涪鋅 70 EQ1 EQ2 Chemical Reaction ML Retrosynthesis prediction Forward Reaction prediction

Slide 100

Slide 100 text

/ 166 *$3F%%⻉㷕⿾䘔ךرؠ؎ٝה涪鋅 70 EQ1 EQ2 Chemical Reaction

Slide 101

Slide 101 text

/ 166 *$3F%%⻉㷕⿾䘔ךرؠ؎ٝה涪鋅 70 Potential Energy Surface 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 E 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 ri AAAChnichVG7TgJBFD2sL8QHqI2JDZFgrMhgUIwV0caShzwSJGR3HXFlX9ldSJD4Aya2UlhpYmH8AD/Axh+w4BOMJSY2Fl6WTYwS8W5m58yZe+6cmSuZqmI7jHV9wtj4xOSUfzowMzs3HwwtLBZso2HJPC8bqmGVJNHmqqLzvKM4Ki+ZFhc1SeVFqb7X3y82uWUrhn7gtExe0cSarhwrsugQlbOqp9VQhMWYG+FhEPdABF6kjdAjDnEEAzIa0MChwyGsQoRNXxlxMJjEVdAmziKkuPsc5wiQtkFZnDJEYuv0r9Gq7LE6rfs1bVct0ykqDYuUYUTZC7tnPfbMHtgr+/yzVtut0ffSolkaaLlZDV4s5z7+VWk0Ozj5Vo307OAY265XhbybLtO/hTzQN886vdxONtpeY7fsjfzfsC57ohvozXf5LsOz1yP8SOSFXowaFP/djmFQ2IjFt2KJTCKS2vVa5ccKVrFO/UgihX2kkaf6NVziCh3BL8SETSE5SBV8nmYJP0JIfQHBBpDP rj EQ1 EQ2 Chemical Reaction

Slide 102

Slide 102 text

/ 166 *$3F%%⻉㷕⿾䘔ךرؠ؎ٝה涪鋅 70 Potential Energy Surface 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 E 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 ri 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 rj EQ1 EQ2 Chemical Reaction

Slide 103

Slide 103 text

/ 166 *$3F%%⻉㷕⿾䘔ךرؠ؎ٝה涪鋅 70 Potential Energy Surface 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 E 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 ri AAAChnichVG7TgJBFD2sL8QHqI2JDZFgrMhgUIwV0caShzwSJGR3HXFlX9ldSJD4Aya2UlhpYmH8AD/Axh+w4BOMJSY2Fl6WTYwS8W5m58yZe+6cmSuZqmI7jHV9wtj4xOSUfzowMzs3HwwtLBZso2HJPC8bqmGVJNHmqqLzvKM4Ki+ZFhc1SeVFqb7X3y82uWUrhn7gtExe0cSarhwrsugQlbOqp9VQhMWYG+FhEPdABF6kjdAjDnEEAzIa0MChwyGsQoRNXxlxMJjEVdAmziKkuPsc5wiQtkFZnDJEYuv0r9Gq7LE6rfs1bVct0ykqDYuUYUTZC7tnPfbMHtgr+/yzVtut0ffSolkaaLlZDV4s5z7+VWk0Ozj5Vo307OAY265XhbybLtO/hTzQN886vdxONtpeY7fsjfzfsC57ohvozXf5LsOz1yP8SOSFXowaFP/djmFQ2IjFt2KJTCKS2vVa5ccKVrFO/UgihX2kkaf6NVziCh3BL8SETSE5SBV8nmYJP0JIfQHBBpDP rj EQ1 EQ2 TS Chemical Reaction

Slide 104

Slide 104 text

/ 166 *$3F%%⻉㷕⿾䘔ךرؠ؎ٝה涪鋅 70 Potential Energy Surface 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 E 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 ri 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 rj EQ1 EQ2 TS Chemical Reaction Energy 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 E ML • Acceleration by ML potential? • Artificial force learning? ML Forces • Scope/network expansion? ML • Fill any gap between theory and experiments (reality) by data?

Slide 105

Slide 105 text

/ 166 71 ⴓ㶨ך䕎׾鋅ג⿾䘔ָ饯ֿ׷״ֲחֲתֻ䬃׃׋׶䒷ְ׋׶כ㷕统דֹ׷˘

Slide 106

Slide 106 text

/ 166 ꆀ㶨⻉㷕堣唒㷕统ך輐さ׾湡䭷׃ג 72

Slide 107

Slide 107 text

/ 166 ׮׃ֻכծءىُٖ٦ءّٝ堣唒㷕统ך輐さ 73 Annu. Rev. Phys. Chem. 71:361–90 (2020) Nat. Rev. Chem. 4: 347–358 (2020) PNAS (2020)

Slide 108

Slide 108 text

/ 166 䩛竲ֹ涸٥鎸〾涸乼⡲ך堣唒㷕统䪮遭׮굲鬨涸ח涪㾜 74 https://uclnlp.github.io/nampi/ Machine intelligence capable of learning complex procedural behavior, inducing (latent) programs, and reasoning with these programs is a key to solving artificial intelligence. Recently, there have been a lot of success stories in the deep learning community related to learning neural networks capable of using trainable memory abstractions. Neural Abstract Machines & Program Induction • Differentiable Neural Computers / Neural Turing Machines (Graves+ 2014) • Memory Networks (Weston+ 2014) • Pointer Networks (Vinyals+ 2015) • Neural Stacks (Grefenstette+ 2015, Joulin+ 2015) • Hierarchical Attentive Memory (Andrychowicz+ 2016) • Neural Program Interpreters (Reed+ 2016) • Neural Programmer (Neelakantan+ 2016) • DeepCoder (Balog+ 2016) : 䩛竲ֹ涸٥鎸〾涸乼⡲׮㷕统דֹ׷فؚٗٓيה׃ג䪔ִ׷״ֲחז׏גֹ׋

Slide 109

Slide 109 text

/ 166 䌓秛غ؎،أךرؠ؎ٝ僇爙涸濼陎堣唒㷕统ך輐さ 75 Variable 1 Variable 2 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 x1 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 x2 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 x1 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 x2 x1 x2 y ML 椚锷⻉㷕זו傀濼ך濼陎׾筨⹛㆞׃גչ搀䠐 ㄂זؿ؍حذ؍ؚٝ穠卓חꤒ׵זְ״ֲחպ ٌرٕך荈歋䏝׾✮׭♳䩛חⵖ秈׃גֶֻ 害欽ך.-ٌرٕדכؿ؍حذ؍ؚٝח⢪ֲ ٌرٕך荈歋䏝ָ넝ֺׅג 6OEFSTQFDJDBUJPOךٔأָؙ넝ֺׅ׷ 植韋׾椚鍑׃׋ְ荈搫猰㷕ⴓꅿדכ㣐㉏겗 Ԩ 堣唒㷕统ٌرٕך䧭⸆ך׋׭חכ㢳ַ׸㼰זַ׸ ٌرٕך䌓秛غ؎،أָ湡⵸ך㉏겗ח黝さ׃גְ ׷䗳銲ָ֮׷ 傀濼זֿהכ㷕统ׅ׷䗳銲ָזְ Ԩ 害欽䚍ך넝ְٌرٕ׾넝ְ然䏝ד㷕统ׅ׷חכ 湱䘔ך花㣐זر٦ةָ䗳銲 㢳ֻך㜥さծꨇ׃ְ 植➿涸.-דכ Ⰵ⸂㢌侧ך㟓㣐 ؗحثٝءؙٝ⻉ הػًٓةٌرٕ侧ך肍㣐⻉ח״׶ ٌرٕך荈歋䏝ָ넝ֺׅ׷˘

Slide 110

Slide 110 text

/ 166 ؚٖ؎نحؙأ剑黝⻉锷椚䱿锷ה窟鎘涸✮庠ך輐さ 76 5IFPSZESJWFO %BUBESJWFO ˖ 㼎韋植韋ך醱꧟⻉ ˖ ءىُٖ٦ءّٝ䪮岀׮醱꧟⻉ ˖ 穗꿀涸ח寸׭׷ػًٓةװⴱ劍⦼ ˖ 害ꟼ侧ծ❛䳔湱ꟼ갪ך鏣鎘 ˖ 㼭؟ٝفٕ٥⡚ؕؐٝزך㉏겗 ˖ 䌓秛غ؎،أךٌرٕؒٝ؝٦س ˖ 㢩䯏ך⡚⥋걾䚍ה䱱稊 ˖ #MBDLCPY䚍٥鍑ꅸ䚍ך㉏겗 ˖ 濼陎ك٦أה锷椚䱿锷 鎸〾"* ךꣲ歲 ˖ ⿑㺘䱿锷װ䱱稊ך鎘皾旗涪 /1㔭ꨇ䚍 ˖ 㣐ꆀر٦ةך濼陎⻉ך㉏겗 ˖ ⵖ秈فؚٗٓىؚٝװ穈さׇ剑黝⻉ ➂䊨濼腉ⴓꅿ ➂䊨濼腉ⴓꅿ ˖ %BUB%SJWFO䩛岀 堣唒㷕统 ה➂꟦ך 锷椚涸䙼罋הך㣐ֹזٍؘحف ˖ %BUBָזְ걄㚖ך䱱稊װչמ׵׭ֹպ ˖ ٌرٕ黝欽眔㔲ה⥋걾䚍٥㸜Ⰻ䚍 倜׋ז倯岀锷פ ر٦ةず⻉ծ垷⦺㷕统ծ锷椚さ䧭ծFUD 邌植㷕统ծٌرٕك٦أ剑黝⻉٥䓼⻉ 㷕统ծًة㷕统ծ欰䧭ٌرٕծFUD ս怴糊٥さ椚锷վ ս䌓秛٥穗꿀锷վ

Slide 111

Slide 111 text

/ 166 Ⰻ⴨䮙٥ⵖ秈➰ֹ穗騟䱱稊̔չ(33.պ 77 ⳿涪朐䡾ַ׵鼂獳〳腉ז&2׾"'*3ד⴨䮙׃ծ倜׋ז&2ָ⳿זֻז׷תד ⱄ䌓涸חֿךⳢ椚׾黝欽 (MPCBM3FBDUJPO3PVUF.BQQJOH ؚٗ٦غٕ⿾䘔穗騟وحؾؚٝ EQ 㹋ꥷחכ⾱㶨玎䏝ך禸דזְהⰋ⴨䮙כ 侧ָ㢳ֺׅג搀椚זךד黝㹅䗳銲ח䘔ׄגⵖ秈 ̔(333.דכ鸞䏝㹀侧遤⴨　簭岀 3$.$ ָⵃ欽〳腉 EQ EQ EQ EQ EQ EQ EQ EQ EQ EQ EQ EQ EQ EQ EQ EQ EQ EQ

Slide 112

Slide 112 text

/ 166 ⿾䘔穗騟طحزٙ٦ؙ 78

Slide 113

Slide 113 text

/ 166 ⿾䘔穗騟طحزٙ٦ؙכ#JPJOGPד侔ղ灇瑔׃גֹ׋㼎韋79 IUUQCJPDIFNJDBMQBUIXBZTDPNNBQ

Slide 114

Slide 114 text

/ 166 ⻉㷕ךمحززؾحؙד֮׷׌ֽדזֻ˘ 80

Slide 115

Slide 115 text

/ 166 ⻉㷕ךمحززؾحؙד֮׷׌ֽדזֻ˘ 81

Slide 116

Slide 116 text

/ 166 堣唒㷕统ⴓꅿךمحززؾحؙד׮֮׷ 82 NeurIPS 2020 ICML 2020, 2021 ICLR 2020, 2021 • Self-Supervised Graph Transformer on Large-Scale Molecular Data • RetroXpert: Decompose Retrosynthesis Prediction Like A Chemist • Reinforced Molecular Optimization with Neighborhood- Controlled Grammars • Autofocused Oracles for Model-based Design • Barking Up the Right Tree: an Approach to Search over Molecule Synthesis DAGs • On the Equivalence of Molecular Graph Convolution and Molecular Wave Function with Poor Basis Set • CogMol: Target-Specific and Selective Drug Design for COVID-19 Using Deep Generative Models • A Graph to Graphs Framework for Retrosynthesis Prediction • Hierarchical Generation of Molecular Graphs using Structural Motifs • Learning to Navigate in Synthetically Accessible Chemical Space Using Reinforcement Learning • Reinforcement Learning for Molecular Design Guided by Quantum Mechanics • Multi-Objective Molecule Generation using Interpretable Substructures • Improving Molecular Design by Stochastic Iterative Target Augmentation • A Generative Model for Molecular Distance Geometry • GraphDF: A Discrete Flow Model for Molecular Graph Generation • An End-to-End Framework for Molecular Conformation Generation via Bilevel Programming • Equivariant message passing for the prediction of tensorial properties and molecular spectra • Learning Gradient Fields for Molecular Conformation Generation • Self-Improved Retrosynthetic Planning • Directional Message Passing for Molecular Graphs • GraphAF: a Flow-based Autoregressive Model for Molecular Graph Generation • Augmenting Genetic Algorithms with Deep Neural Networks for Exploring the Chemical Space • A Fair Comparison of Graph Neural Networks for Graph Classification • MARS: Markov Molecular Sampling for Multi-objective Drug Discovery • Practical Massively Parallel Monte-Carlo Tree Search Applied to Molecular Design • Learning Neural Generative Dynamics for Molecular Conformation Generation • Conformation-Guided Molecular Representation with Hamiltonian Neural Networks • Symmetry-Aware Actor-Critic for 3D Molecular Design

Slide 117

Slide 117 text

/ 166 ➙傈ךذ٦و 83 ˖ 荈䊹稱➜ 堣唒㷕统ה荈搫猰㷕ך㞮歲 ˖ 堣唒㷕统הכ倜׃ְفؚٗٓىؚٝך倯岀 ˖ 堣唒㷕统㾊כ♧⡤⡦ָ嚂׃ְךַ ˖ ⴓ㶨ך邌植ה堣唒㷕统 ˖ ؚٖ؎نحؙأ剑黝⻉ 怴糊䌓秛 锷椚㷕ה窟鎘㷕ך輐さ ˖ 荈搫猰㷕灇瑔ד堣唒㷕统׾⢪ֲֶהׅ׷ה䗳׆עאַ׷劤䔲חꨇ׃ְ㉏겗 ˖ ر٦ةٌرؚٔٝה✮庠،ٕ؞ٔؤي 5IF5XP$VMUVSFT ˖ ✮庠ַ椚鍑ַ3BTIPNPO⸬卓 6OEFSTQFDJDBUJPO 鍑ꅸ㢳圫䚍 ˖ ➂꟦ך钠濼غ؎،أח歋勻ׅ׷㉏겗⟎铡ծ㣟侁ծ䧭⸆غ؎،أծFUD ˖ 堣唒㷕统ַ׵堣唒涪鋅פ ˖ չ涪鋅պչ椚鍑պך麣瘡כさ椚⻉דֹ׷ךַ荈⹛⻉דֹ׷ךַ

Slide 118

Slide 118 text

/ 166 ت٦ؙ؟؎سפ״ֲֿ׉ 84 Ԩ ֿ׸תדך鑧כ⚺חչꆀ㶨⻉㷕鎘皾ח״׷ر٦ةպדؙٔ٦ٝז⚅歲 錁庠ظ؎ؤָזְ٥⳿⸂׾䖤׷ךח䗳銲⼧ⴓזⰅ⸂䞔㜠ָⴓַ׏גְ׷٥ ְ׹ְ׹זؔ٦فٝر٦ةָⵃ欽דֹ׷٥FUD Ԩ 植㹋כא׵ְ˘ ˖ 錁庠ظ؎ؤָ֮׶暟椚涸醱醡ָ䗳銲 ✳䏝庠׷ה⦼ָ殯ז׷倯ָ兛鸐 ˖ 椚锷鎘皾ח《׶Ⰵ׸׵׸גזְ搀侧ך❛窃㔓㶨װ㢩✉㔓㶨ך䕦갟 ˖ 醱꧟禸דכⰅ⸂㢌侧ח⡦׾Ⰵ׸׷ץֹזךַָ♶僇הְֲآٖٝو ̔Ⰵ⳿⸂ꟼ⤘ך堣䎷ָⴓַ׵זְַ׵堣唒㷕统׾⢪ְ׋ְךח խ䗳銲ז䞔㜠׾Ⰵ⸂חⰅ׸זְה堣唒㷕统חכ亻⡂湱ꟼ׃ַ鋅ִזְ ˖ ׉׮׉׮鎘庠٥ⵖ䖴דֹזְ׋ֻׁ׿ךغحؙؚٓؐٝس㔓㶨ָ֮׷ ˖ ➂꟦ָ㹋꿀׾鎘歗ׅ׷ה䖤׵׸׷ر٦ةכ䌢חغ؎،أ׾ろ׬ ̔չ葺ְㅷ颵ךպ䗳銲⼧ⴓז鋅劤⢽׾⡲׷ךכ劤䔲חꨇ׃ְ

Slide 119

Slide 119 text

/ 166 ت٦ؙ؟؎سַ׵ך5BLF)PNFًحإ٦آ 85 荈搫猰㷕ⴓꅿדךⵃ崞欽כ.-ך䪮遭灇熊׌ֽדכ䧭⸆׃זְկ ⴓꅿ㼔Ꟍ㹺הך⼿⫴ָ䗳銲♶〳妀

Slide 120

Slide 120 text

/ 166 ت٦ؙ؟؎سַ׵ך5BLF)PNFًحإ٦آ 85 ˖ .-ָוְֲֲ䪮遭זךַ.-ך暴䚍הꣲ歲׾姻׃ֻ䪾䳢ׅ׷ 荈搫猰㷕ⴓꅿדךⵃ崞欽כ.-ך䪮遭灇熊׌ֽדכ䧭⸆׃זְկ ⴓꅿ㼔Ꟍ㹺הך⼿⫴ָ䗳銲♶〳妀

Slide 121

Slide 121 text

/ 166 ت٦ؙ؟؎سַ׵ך5BLF)PNFًحإ٦آ 85 ˖ .-ָוְֲֲ䪮遭זךַ.-ך暴䚍הꣲ歲׾姻׃ֻ䪾䳢ׅ׷ ˖ չر٦ةך　꧊鎘歗 㹋꿀鎘歗 הㅷ颵⥂鏾ծ黝欽眔㔲ך椚鍑պ ָˑEBUBESJWFO˒ך䗰茖ד֮׷ֿה׾ְא׮䗰ח 荈搫猰㷕ⴓꅿדךⵃ崞欽כ.-ך䪮遭灇熊׌ֽדכ䧭⸆׃זְկ ⴓꅿ㼔Ꟍ㹺הך⼿⫴ָ䗳銲♶〳妀

Slide 122

Slide 122 text

/ 166 ت٦ؙ؟؎سַ׵ך5BLF)PNFًحإ٦آ 85 ˖ .-ָוְֲֲ䪮遭זךַ.-ך暴䚍הꣲ歲׾姻׃ֻ䪾䳢ׅ׷ ˖ չر٦ةך　꧊鎘歗 㹋꿀鎘歗 הㅷ颵⥂鏾ծ黝欽眔㔲ך椚鍑պ ָˑEBUBESJWFO˒ך䗰茖ד֮׷ֿה׾ְא׮䗰ח ˖ չ䱱稊պָ湡涸ז׵.-ך卓׋ׅ䕵ⶴכֻ֮תד♧鿇ה䗰䖤׷ " 㼔Ꟍ㹺הך⼿⫴ծⴓꅿך㼔Ꟍ濼陎ח撑׵׃׋嗚鏾٥鍑ꅸ " ءىُٖ٦ءّٝ٥㹋꿀荈⹛⻉٥锷椚䱿锷הך輐さ 荈搫猰㷕ⴓꅿדךⵃ崞欽כ.-ך䪮遭灇熊׌ֽדכ䧭⸆׃זְկ ⴓꅿ㼔Ꟍ㹺הך⼿⫴ָ䗳銲♶〳妀

Slide 123

Slide 123 text

/ 166 5IF5XP$VMUVSFT 86 https://projecteuclid.org/euclid.ss/1009213726 (Open Access)

Slide 124

Slide 124 text

/ 166 -FP#SFJNBO ָ㥨ֹ 87 ˖ $"35 $MBTTJDBUJPOBOE3FHSFTTJPO5SFFT 1*.1-& ˖ 3BOEPN'PSFTU ˖ "SDJOH BLB#PPTUJOH ˖ #BHHJOH 1BTUJOH ˖ "$& "MUFSOBUJWF$POEJUJPOBM&YQFDUBUJPOT ˖ 4UBDLFE(FOFSBMJ[BUJPO BLB4UBDLJOH#MFOEJOH ˖ /POOFHBUJWF(BSSPUF -"440ך⵸魦GPS4VCTFU㔐䌓 ˖ *OTUBCJMJUZ4UBCJMJ[BUJPOJO.PEFM4FMFDUJPO ˖ 4IBOOPO.D.JMMBO#SFJNBO5IFPSFN 恷鵚瘝ⴓⶴ䚍 ˖ ,FMMZ#SFJNBO4USBUFHZ 剑黝ז㹀嫰⢽䨌殛 ˖ 6$#FSLFMFZせ钙侄䱇 ˖ 4*(,%%*OOPWBUJPO"XBSE ˖ 1SPCBCJMJUZ5IFPSJTU̔$POTVMUBOU̔4UBUJTUJUJBO *GTUBUJTUJDTJTBOBQQMJFEFMEBOEOPUBNJOPSCSBODIPGNBUIFNBUJDT UIFOPGUIFQVCMJTIFEQBQFSTBSFVTFMFTTFYFSDJTFT 3FFDUJPOTBGUFSSFGFSFFJOHQBQFSTGPS/*14˒ 5IF.BUIFNBUJDTPG(FOFSBMJ[BUJPO &E%)8PMQFSU https://en.wikipedia.org/wiki/File:Leo_Breiman.jpg 侧㷕ך1I% 6$#FSLFMFZ ̔侧㷕猰UFOVSF侄㆞ 6$-" ̔窟鎘؝ٝ؟ٕةٝز 䎃 ̔窟鎘㷕猰侄䱇 6$#FSLFMFZ

Slide 125

Slide 125 text

/ 166 -FP#SFJNBO ָ㥨ֹ 88 "OOBMTPG"QQMJFE4UBUJTUJDT 7PM /P %FDFNCFS ח#SFJNBOך 鷄䞄暴꧊ָ֮׶ծ葿ղזꟼ⤘罏ָ䙼ְ⳿ װ娖〷׾铂׏גְתׅ ؿ؋ٝ䗳鋅 ׮׍׹׿'SJFENBO 0MTIFO 4UPOF׮

Slide 126

Slide 126 text

/ 166 89

Slide 127

Slide 127 text

/ 166 90

Slide 128

Slide 128 text

/ 166 91

Slide 129

Slide 129 text

/ 166 5IF5XP$VMUVSFT 92

Slide 130

Slide 130 text

/ 166 5IF5XP$VMUVSFT窟鎘㷕WT堣唒㷕统 93 The Data Modeling Culture The Algorithmic Modeling Culture ⠗窟涸ז 窟鎘㷕 堣唒㷕统 "Generative Models" "Discriminative Models" vs vs "Explanatory Modeling" "Predictive Modeling" vs

Slide 131

Slide 131 text

/ 166 -FP#SFJNBOךٖحأٝ 94

Slide 132

Slide 132 text

/ 166 -FP#SFJNBOךٖحأٝ 95 ➙ד׮ֿך挿כת׌葿ղה0QFOז㉏겗׾㶱׿ד灇瑔ׁ׸גְ׷ Ӎ "Rashomon"葺ְٌرٕך㢳ꅾ䚍 ꬊ♧䠐䚍 ず玎䏝ך葺ְ✮庠礵䏝׾䭯אⰋֻ殯ז׷ٌرָٕ׋ֻׁ׿㶷㖈ׅ׷ Ӎ "Occam"ٌرٕך鍑ꅸ䚍ה✮庠礵䏝ך؝ٝؿؙٔز ٌرٕךءٝفׁٕ 鍑ꅸ䚍 ה✮庠礵䏝ך⚕甧כהג׮ꨇ׃ְ Ӎ "Bellman"넝如⯋ر٦ةָ䒷ֹ饯ֿًׅٔحزהرًٔحز 넝如⯋ז邌植 ꟼ⤘׃׉ֲזדֹ׷׌ֽ㢳ֻך㢌ꆀ ׾䪔ֲץֹזךַ ⠗窟涸ז窟鎘㷕ך״ֲח佄ꂁ涸ז㼰侧ך㢌ꆀ׾嗚鎢׃ⴓ匿ׅץֹזךַ ̔ؗحثٝءؙٝ㔐䌓 䙼ְאֻ㢌侧Ⰻ鿇Ⰵ׶ٌرٕ ٥暴䗙ꆀ ؒٝآٔ،ؚٔٝה亻⡂湱ꟼ٥3BTIPNPO⸬卓ך㟓㣐ٔأؙ 3BTIPNPO 0DDBN #FMMNBO

Slide 133

Slide 133 text

/ 166 猘׋׍ך蕱ꡙ♶㖱♧禸鍗㯭ךرؠ؎ٝה䱱稊 96 https://en.wikipedia.org/wiki/Heterogeneous_catalysis 㼎韋䊨噟さ䧭٥䱖ؖأ崭⻉٥ًةٝ鯄䳔זו㔿⡤鍗㯭邌꬗♳ך孡湱⿾䘔

Slide 134

Slide 134 text

/ 166 ♶㖱♧禸鍗㯭灇瑔דך堣唒㷕统ך崞欽 97 • Mine+ ChemCatChem. 2021. • Toyao+, ACS Catalysis. 2020. (Review) • Liu+, The Journal of Physical Chemistry C. 2020. • Suzuki+ ChemCatChem. 2019. (Front Cover) • Kamachi+ The Journal of Physical Chemistry C. 2019. • Hinuma+ The Journal of Physical Chemistry C. 2018. • Toyao+, The Journal of Physical Chemistry C. 2018 • Takigawa+ RSC Advances. 2016. +45$3&45ꬠ倜勞俱Ꟛ涪 稢ꅿ걄㚖 鍗㯭؎ٝؿؓوذ؍ؙأךⶼ䧭ך׋׭ך㹋꿀٥椚锷٥ر٦ة猰㷕灇瑔 幠宏灇♧ 넝虊加麦 둷㾊㽵ꥐ ⵸ꅿ犣 넝㽵㛇〷 䂹溪⛲ ꈿ加䣒➜ ֿך竲ֹ暟ך 灇瑔׾稱➜

Slide 135

Slide 135 text

/ 166 ⿫罋5PZBP "$4$BUBMZTJT 3FWJFX 98 • This is an excellent review on a very timely subject, which is highly suitable for ACS Catalysis. … I don't usually recommend that papers should be accepted "as is", but in this case I don't see the need for changes. • I will certainly recommend it to my group and my students when it is published. • The manuscript gives an excellent overview in the field of machine learning especially with regard to heterogeneous catalysis and I would highly recommend the article for the publication in ACS Catalysis. • This is one of the best reviews for catalyst informatics that the reviewer has read. In particular, the chapter 2 delivers a very good tutorial, which is concisely and professionally written. Review Comments 畍ָ堣唒㷕统ךِ٦ؠؖ؎س 侧䒭ז׃ חז׏גְתׅ

Slide 136

Slide 136 text

/ 166 ㉏겗ךꨇ׃ׁך然钠 99 孡湱 ⿾䘔暟 㔿湱 鍗㯭 չ㔿⡤鍗㯭邌꬗♳ך孡湱⿾䘔 醱꧟禸 պך椚鍑כ׉׮׉׮慧يؤ 䬐䭯ꆃ㾩شظ磛㶨ך邌꬗

Slide 137

Slide 137 text

/ 166 ㉏겗ךꨇ׃ׁך然钠 99 孡湱 ⿾䘔暟 㔿湱 鍗㯭 չ㔿⡤鍗㯭邌꬗♳ך孡湱⿾䘔 醱꧟禸 պך椚鍑כ׉׮׉׮慧يؤ ェ滠 䬐䭯ꆃ㾩شظ磛㶨ך邌꬗

Slide 138

Slide 138 text

/ 166 ㉏겗ךꨇ׃ׁך然钠 99 孡湱 ⿾䘔暟 㔿湱 鍗㯭 չ㔿⡤鍗㯭邌꬗♳ך孡湱⿾䘔 醱꧟禸 պך椚鍑כ׉׮׉׮慧يؤ ェ滠 䭁侔 䬐䭯ꆃ㾩شظ磛㶨ך邌꬗

Slide 139

Slide 139 text

/ 166 ㉏겗ךꨇ׃ׁך然钠 99 孡湱 ⿾䘔暟 㔿湱 鍗㯭 չ㔿⡤鍗㯭邌꬗♳ך孡湱⿾䘔 醱꧟禸 պך椚鍑כ׉׮׉׮慧يؤ ェ滠 䭁侔 鍑ꨄ 䬐䭯ꆃ㾩شظ磛㶨ך邌꬗

Slide 140

Slide 140 text

/ 166 ㉏겗ךꨇ׃ׁך然钠 99 孡湱 ⿾䘔暟 㔿湱 鍗㯭 չ㔿⡤鍗㯭邌꬗♳ך孡湱⿾䘔 醱꧟禸 պך椚鍑כ׉׮׉׮慧يؤ ェ滠 䭁侔 鍑ꨄ ⿾䘔 䬐䭯ꆃ㾩شظ磛㶨ך邌꬗

Slide 141

Slide 141 text

/ 166 ㉏겗ךꨇ׃ׁך然钠 99 孡湱 ⿾䘔暟 㔿湱 鍗㯭 չ㔿⡤鍗㯭邌꬗♳ך孡湱⿾䘔 醱꧟禸 պך椚鍑כ׉׮׉׮慧يؤ ェ滠 䭁侔 鍑ꨄ ⿾䘔 膴ꨄ 䬐䭯ꆃ㾩شظ磛㶨ך邌꬗

Slide 142

Slide 142 text

/ 166 ㉏겗ךꨇ׃ׁך然钠 99 孡湱 ⿾䘔暟 㔿湱 鍗㯭 չ㔿⡤鍗㯭邌꬗♳ך孡湱⿾䘔 醱꧟禸 պך椚鍑כ׉׮׉׮慧يؤ ェ滠 䭁侔 鍑ꨄ ⿾䘔 膴ꨄ 䬐䭯ꆃ㾩شظ磛㶨ך邌꬗ ˖ 鍗㯭穈䧭 ˖ ؟؎ؤװ䕎朐 磛㶨ך⡲׶倯 ˖ 邌꬗ך⳽⳻ ˖ 庛䏝װ㖇⸂ 㢳㔓㶨ָꟼ׻׷

Slide 143

Slide 143 text

/ 166 ㉏겗ךꨇ׃ׁך然钠 99 孡湱 ⿾䘔暟 㔿湱 鍗㯭 չ㔿⡤鍗㯭邌꬗♳ך孡湱⿾䘔 醱꧟禸 պך椚鍑כ׉׮׉׮慧يؤ ˑ邌꬗猰㷕˒ ׉׮׉׮չ邌꬗պָ䝤눤涸זꨇ׃ׁ God made the bulk; the surface was invented by the devil ύ΢Ϧେઌੜ ェ滠 䭁侔 鍑ꨄ ⿾䘔 膴ꨄ 䬐䭯ꆃ㾩شظ磛㶨ך邌꬗ ˖ 鍗㯭穈䧭 ˖ ؟؎ؤװ䕎朐 磛㶨ך⡲׶倯 ˖ 邌꬗ך⳽⳻ ˖ 庛䏝װ㖇⸂ 㢳㔓㶨ָꟼ׻׷

Slide 144

Slide 144 text

/ 166 㹋꿀ָ⚺㼪׃ג葺ְ鍗㯭ָ鋅אַ׏גֹ׋ֿהכ끅殯涸 100 ְתתדך 濼鋅װر٦ة ⡲噟⟎铡̔㹋꿀 椚锷ٌرٕ̔鎘皾 穠卓ך然钠ה 怴糊涸ז嗚鏾 ˑ穗꿀ה⹞˒ ٗإٓؕءؚٝؗٝ ⟎铡怴糊岀

Slide 145

Slide 145 text

/ 166 㹋꿀ָ⚺㼪׃ג葺ְ鍗㯭ָ鋅אַ׏גֹ׋ֿהכ끅殯涸 100 ְתתדך 濼鋅װر٦ة ⡲噟⟎铡̔㹋꿀 椚锷ٌرٕ̔鎘皾 穠卓ך然钠ה 怴糊涸ז嗚鏾 ˑ穗꿀ה⹞˒ ٗإٓؕءؚٝؗٝ ⟎铡怴糊岀 • Genius is 1% inspiration and 99% perspiration. • There is no substitute for hard work. • I have not failed. I've just found 10,000 ways that won't work. Τδιϯେઌੜ PSˑؒآاٝ涸ז˒穗꿀锷

Slide 146

Slide 146 text

/ 166 㹋꿀ָ⚺㼪׃ג葺ְ鍗㯭ָ鋅אַ׏גֹ׋ֿהכ끅殯涸 100 ְתתדך 濼鋅װر٦ة ⡲噟⟎铡̔㹋꿀 椚锷ٌرٕ̔鎘皾 穠卓ך然钠ה 怴糊涸ז嗚鏾 ˑ穗꿀ה⹞˒ ٗإٓؕءؚٝؗٝ ⟎铡怴糊岀 銲秈ׅ׷הչ⸕⸂֮׷ך׫החַֻ׋ֻׁ׿ָ׿ל׸պה鎉׏גְ׷կ ֶ̔ꆃך䫎Ⰵ➂嵲䨌遭 هأسؙװ㷕欰ך麓ꃎז⸤⫴ ד(0 • Genius is 1% inspiration and 99% perspiration. • There is no substitute for hard work. • I have not failed. I've just found 10,000 ways that won't work. Τδιϯେઌੜ PSˑؒآاٝ涸ז˒穗꿀锷

Slide 147

Slide 147 text

/ 166 ׉׸ד׮չ涪鋅պכ׮ךׅ׀ֻٖ،؎كٝزד֮׷ 101 䟝㹀דֹ׷չ鍗㯭㹋꿀勴⟝倯岀պך侧כ㣓俑㷕涸ח䊬㣐 # 剣ꣲך儗꟦٥؝أز׾欰ֹ׷猘׋׍ָ鑐ׇ׷ךכק׿ך♧鿇 # 醱꧟⻉ׅ׷ص٦ؤ׾⿾僥׃׋稆兦׵׃ְ歗劍涸ז鍗㯭ָ 鋅אַ׷然桦כ椚㾄♳כ窫劄涸ח⡚ְ˘כ׆ ˑ"OFFEMFJOBIBZTUBDL˒

Slide 148

Slide 148 text

/ 166 את׶չإٖٝر؍ؾذ؍͒⩐搫ך䎊麊պ 102 䎊麊כ彊⪒ׁ׸׋׮ךח׌ֽ꣬׶׷ ˖ ⡦׾㹋꿀ׅ׷ַ ⟎铡䕎䧭 כסאֲ㸣Ⰻחٓٝتي ְֹ֮׋׶ ל׏׋׶ דכזְկ ˖ 穗꿀ה⹞չ灇瑔罏ךإٝأպװչ臾ך鋅ׇ䨽պ ˖ ⮚׸׋㹋꿀猰㷕罏ךչ⹞ؾُ٦ة 穗꿀ה⹞ պכٓٝتيדכ זֻ⡦׵ַך䭷ぢ䚍׾䭯א

Slide 149

Slide 149 text

/ 166 את׶չإٖٝر؍ؾذ؍͒⩐搫ך䎊麊պ 102 䎊麊כ彊⪒ׁ׸׋׮ךח׌ֽ꣬׶׷ ˖ ⡦׾㹋꿀ׅ׷ַ ⟎铡䕎䧭 כסאֲ㸣Ⰻחٓٝتي ְֹ֮׋׶ ל׏׋׶ דכזְկ ˖ 穗꿀ה⹞չ灇瑔罏ךإٝأպװչ臾ך鋅ׇ䨽պ ˖ ⮚׸׋㹋꿀猰㷕罏ךչ⹞ؾُ٦ة 穗꿀ה⹞ պכٓٝتيדכ זֻ⡦׵ַך䭷ぢ䚍׾䭯א ֿך֮׋׶חˑEBUBESJWFO˒ָ顀柃דֹ׷⡭㖑ָ֮׷ ㉏겗㹋ꥷחכر٦ة⻉דֹזְ䞔㜠ָקה׿וזךד ر٦ة⻉דֹ׷䞔㜠ך⚥ךչוְֲֲر٦ةדESJWFׅ׷ךַպ

Slide 150

Slide 150 text

/ 166 וְֲֲر٦ةד⟎铡䕎䧭٥㹋꿀鎘歗׾ESJWFדֹ׷ַ 103 ˖ ⮚׸׋㹋꿀猰㷕罏ח➙תדךⰋ➂欰דⰅ⸂ׁ׸׋䞔㜠כ花㣐 ر٦ة⻉ׁ׸זְ䞔㜠ָקה׿ו ˖ ر٦ة꽀⹛䩛חⰅ׷ر٦ةַ׵鵚⡂涸ח鶕׷׃ַזְָ ➂꟦ך钠濼ꣲ歲ךⵖ秈װ䙼ְ鴥׫ח״׷勲簂ַ׵荈歋חז׷ ְתתדך 濼鋅װر٦ة ⡲噟⟎铡̔㹋꿀 椚锷ٌرٕ̔鎘皾 穠卓ך然钠ה 怴糊涸ז嗚鏾 ⟎铡䕎䧭٥㹋꿀鎘歗ח 堣唒㷕统׾⢪ֲֶ ٗإٓؕءؚٝؗٝ ̞➂꟦כ㢳侧ך㔓㶨ך醱꧟ז㢳如⯋湱ꟼ׾䪾䳢דֹזְ

Slide 151

Slide 151 text

/ 166 堣唒㷕统׾וֲ崞欽ׅ׷ַ 104 堣唒㷕统׾崞欽ׅ׷׋׭חכ 堣唒㷕统ٌرٕ׾鎮箺ׅ׷׋׭ךչر٦ةպ׾וֲׅ׷ַָ꒲ ˖ 俑柃ַ׵꧊׭׋㹋ꥷך㹋꿀ر٦ة㜠デ׾⢪ֲ ˖ ٓنד㹋꿀׃ג襳琎׃׋ر٦ة׾⢪ֲ ˖ ءىُٖ٦ءّٝ 鎘皾⻉㷕 ד襳琎׃׋ر٦ة׾⢪ֲ ˖ ♳鎸אⰋ鿇⢪ֲ ˖ 㹋꿀鎘庠堣㐻חإٝ؟٦׾אֽתֻ׶ծ㹋꿀׃גְ׷הֿ׹׮ ؽرؔꐮ歗׃ծ㹋꿀罏ך걧ח׮ًؕٓאֽג㹋꿀罏鋔ꅿ׮ؽر ؔꐮ歗׃ծ㹋꿀罏ך⡤ח⹛⡲إٝ؟אֽג鎸ꐮ׃ծ㹋꿀ظ٦ز ׮أٍؗٝ׃ծ֮׵ײ׷ꟼ鸬锷俑װ侄猰剅׮Ⰻ鿇ꨵ㶨⻉׃ծ˘

Slide 152

Slide 152 text

/ 166 俑柃ַ׵꧊׭׋㹋ꥷך㹋꿀ر٦ة㜠デ׾⢪ֲ 105 鎘皾⻉㷕כ굲鬨涸涪㾜׾鹴־גְ׷ָ椚锷ה植㹋ךٍؘحفכ ְאתד׮㶷㖈ׅ׷կ ˖ 㹋꿀勴⟝װفٗإأ勴⟝זוך㔓㶨 ˖ 椚锷ךꥷח镘׭׋ PSׂ׏ֻ׶鵚⡂׃׋ 稢ַֺׅ׷搀侧ך銲㔓 ˖ ٌٕחכ،نؖسٗ㹀侧 – ⦐ך銲稆ָ劤䔲כ֮׷ 俑柃ַ׵꧊׭׋㹋ꥷך㹋꿀ر٦ة㜠デ׾⢪ֲ չ麓⿠ח㜠デׁ׸׋植㹋׾鋅ג׫״ֲպ 植㹋׾ׅץגٌرٕ⻉ׅ׷ֿהכ♶〳腉 ̔׉ך㹀纏ַ׵׃גˑٌرٕ˒הכ⡦׵ַך䰍韋װ鵚⡂׾ろ׬կ

Slide 153

Slide 153 text

/ 166 俑柃ַ׵꧊׭׋㹋ꥷך㹋꿀ر٦ة㜠デ׾⢪ֲ 106 ⯋稆穈䧭 㹋꿀勴⟝ 　桦 鼅䫛䚍 㼎韋כًةٝךꃐ⻉ؕحفؚٔٝ⿾䘔ծ湡涸㢌侧כ$ 　桦 䖞勻灇瑔 ;BWZBMPWBFUBM ח״׷䎃⟃⵸ך⢽ח ։䎃ך倜׋ז⢽׾⸇ִ⢽חתד䭁⯍ 堣唒㷕统ד✮庠

Slide 154

Slide 154 text

/ 166 ֿךהֹ堣唒㷕统ח⡦ָ䗳銲ַ 107 ֿך邌ر٦ةד堣唒㷕统ٌرٕ׾鎮箺׃✮庠ׅ׸ל葺ְ 堣唒㷕统ٌرٕ ˖ ⯋稆穈䧭嫰 ˖ 㹋꿀勴⟝ ˖ 　桦 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 x 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 y Ⰵ⸂ ⳿⸂ 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 x Ⰵ⸂ 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 y ⳿⸂ 鎮箺ر٦ة

Slide 155

Slide 155 text

/ 166 ֿךהֹ堣唒㷕统ח⡦ָ䗳銲ַ 107 ֿך邌ر٦ةד堣唒㷕统ٌرٕ׾鎮箺׃✮庠ׅ׸ל葺ְ 堣唒㷕统ٌرٕ ˖ ⯋稆穈䧭嫰 ˖ 㹋꿀勴⟝ ˖ 　桦 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 x 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 y Ⰵ⸂ ⳿⸂ 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 x Ⰵ⸂ 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 y ⳿⸂ 鎮箺ر٦ة

Slide 156

Slide 156 text

/ 166 ֿךהֹ堣唒㷕统ח⡦ָ䗳銲ַ 107 ֿך邌ر٦ةד堣唒㷕统ٌرٕ׾鎮箺׃✮庠ׅ׸ל葺ְ 堣唒㷕统ٌرٕ ˖ ⯋稆穈䧭嫰 ˖ 㹋꿀勴⟝ ˖ 　桦 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 x AAAChHichVHLSsNAFD1GrbW+qm4EN8WiuJBy6xsXUnTjsg9rC1VKEqc1mCYhSQu1+AO6VVy4UnAhfoAf4MYfcNFPEJcKblx4mwZEi3rDZM6cuefOmbmKpWuOS9Tskrp7egN9wf7QwODQ8Eh4dGzHMau2KrKqqZt2XpEdoWuGyLqaq4u8ZQu5ougipxxutvZzNWE7mmlsu3VL7FXksqGVNFV2mUrVi+EoxciLSCeI+yAKP5Jm+B672IcJFVVUIGDAZaxDhsNfAXEQLOb20GDOZqR5+wLHCLG2ylmCM2RmD/lf5lXBZw1et2o6nlrlU3QeNisjmKYnuqVXeqQ7eqaPX2s1vBotL3WelbZWWMWRk4nM+7+qCs8uDr5Uf3p2UcKq51Vj75bHtG6htvW1o4vXzFp6ujFD1/TC/q+oSQ98A6P2pt6kRPryDz8Ke+EX4wbFf7ajE+zMx+LLMUotRhMbfquCmMQUZrkfK0hgC0lkub7AKc5wLgWkOWlBWmqnSl2+ZhzfQlr/BNcbj/U= y Ⰵ⸂ ⳿⸂ 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 x Ⰵ⸂ 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 y ⳿⸂ 傀濼ך剑㣐　桦 鎮箺ر٦ة

Slide 157

Slide 157 text

/ 166 ֿךהֹ堣唒㷕统ח⡦ָ䗳銲ַ 107 ֿך邌ر٦ةד堣唒㷕统ٌرٕ׾鎮箺׃✮庠ׅ׸ל葺ְ 堣唒㷕统ٌرٕ ˖ ⯋稆穈䧭嫰 ˖ 㹋꿀勴⟝ ˖ 　桦 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 x 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 y Ⰵ⸂ ⳿⸂ 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 x Ⰵ⸂ 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 y ⳿⸂ 傀濼ך剑㣐　桦 如挿 鎮箺ر٦ة

Slide 158

Slide 158 text

/ 166 濼׶׋ְֿהח䘔ׄג堣唒㷕统ך倯岀װ岣䠐挿ָ㢌׻׷ 108 ˖ 　桦խָ넝ְ鍗㯭ה⡚ְ鍗㯭ך麩ְ׾鋉㹀ׅ׷㔓㶨כ⡦ ˖ 葺ְ　桦ָ䖤׵׸׷劢涪鋅ך鍗㯭כוך֮׋׶ח֮׷ 剑㣐　桦ָ䖤׵׸׋ךワ鴟זך׌׹ֲַ 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 x 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 x Ⰵ⸂ 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 y ⳿⸂ 傀濼ך剑㣐　桦 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 x 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 y 湡涸䱱稊 傀濼ך鍗㯭״׶葺ְ鍗㯭׾鋅אֽ׋ְ

Slide 159

Slide 159 text

/ 166 ֿךהֹ堣唒㷕统ח⡦ָ䗳銲ַ 109 ֿך邌ر٦ةד堣唒㷕统ٌرٕ׾鎮箺׃✮庠ׅ׸ל葺ְ 堣唒㷕统ٌرٕ ˖ ⯋稆穈䧭嫰 ˖ 㹋꿀勴⟝ ˖ 　桦 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 x AAAChHichVHLSsNAFD1GrbW+qm4EN8WiuJBy6xsXUnTjsg9rC1VKEqc1mCYhSQu1+AO6VVy4UnAhfoAf4MYfcNFPEJcKblx4mwZEi3rDZM6cuefOmbmKpWuOS9Tskrp7egN9wf7QwODQ8Eh4dGzHMau2KrKqqZt2XpEdoWuGyLqaq4u8ZQu5ougipxxutvZzNWE7mmlsu3VL7FXksqGVNFV2mUrVi+EoxciLSCeI+yAKP5Jm+B672IcJFVVUIGDAZaxDhsNfAXEQLOb20GDOZqR5+wLHCLG2ylmCM2RmD/lf5lXBZw1et2o6nlrlU3QeNisjmKYnuqVXeqQ7eqaPX2s1vBotL3WelbZWWMWRk4nM+7+qCs8uDr5Uf3p2UcKq51Vj75bHtG6htvW1o4vXzFp6ujFD1/TC/q+oSQ98A6P2pt6kRPryDz8Ke+EX4wbFf7ajE+zMx+LLMUotRhMbfquCmMQUZrkfK0hgC0lkub7AKc5wLgWkOWlBWmqnSl2+ZhzfQlr/BNcbj/U= y Ⰵ⸂ ⳿⸂ LinearRegression() խ׾㣐ֹֻׅ׸ל 葺ְהְֲ閯ך 爙㇗׃ַֻ׸זְ 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 x Ⰵ⸂ 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 y ⳿⸂ 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 x 6OEFSU

Slide 160

Slide 160 text

/ 166 ֿךהֹ堣唒㷕统ח⡦ָ䗳銲ַ 110 ֿך邌ر٦ةד堣唒㷕统ٌرٕ׾鎮箺׃✮庠ׅ׸ל葺ְ 堣唒㷕统ٌرٕ ˖ ⯋稆穈䧭嫰 ˖ 㹋꿀勴⟝ ˖ 　桦 AAACiXichVFNLwNRFD0dX1VfxUZi02iIVXNHBOmq0Y1lP7QkiMyMh9H5ysy0UU3/gJWdYEViIX6AH2DjD1j0J4gliY2F2+kkguBO3rzzzrvnvvPeVR1D93yiVkTq6u7p7Yv2xwYGh4ZH4qNjZc+uupooabZhu+uq4glDt0TJ131DrDuuUEzVEGtqJdveX6sJ19Nta9WvO2LLVPYsfVfXFJ+p8qZqNg6b2/EkpSiIxE8ghyCJMHJ2/A6b2IENDVWYELDgMzagwONvAzIIDnNbaDDnMtKDfYEmYqytcpbgDIXZCv/3eLURshav2zW9QK3xKQYPl5UJTNMj3dALPdAtPdH7r7UaQY22lzrPakcrnO2R44ni278qk2cf+5+qPz372MVS4FVn707AtG+hdfS1o9OXYrow3ZihK3pm/5fUonu+gVV71a7zonDxhx+VvfCLcYPk7+34CcpzKXkhRfn5ZGY5bFUUk5jCLPdjERmsIIcS1z/ACc5wLg1IsrQkpTupUiTUjONLSNkPIVKSSQ== x AAAChHichVHLSsNAFD1GrbW+qm4EN8WiuJBy6xsXUnTjsg9rC1VKEqc1mCYhSQu1+AO6VVy4UnAhfoAf4MYfcNFPEJcKblx4mwZEi3rDZM6cuefOmbmKpWuOS9Tskrp7egN9wf7QwODQ8Eh4dGzHMau2KrKqqZt2XpEdoWuGyLqaq4u8ZQu5ougipxxutvZzNWE7mmlsu3VL7FXksqGVNFV2mUrVi+EoxciLSCeI+yAKP5Jm+B672IcJFVVUIGDAZaxDhsNfAXEQLOb20GDOZqR5+wLHCLG2ylmCM2RmD/lf5lXBZw1et2o6nlrlU3QeNisjmKYnuqVXeqQ7eqaPX2s1vBotL3WelbZWWMWRk4nM+7+qCs8uDr5Uf3p2UcKq51Vj75bHtG6htvW1o4vXzFp6ujFD1/TC/q+oSQ98A6P2pt6kRPryDz8Ke+EX4wbFf7ajE+zMx+LLMUotRhMbfquCmMQUZrkfK0hgC0lkub7AKc5wLgWkOWlBWmqnSl2+ZhzfQlr/BNcbj/U= y Ⰵ⸂ ⳿⸂ MLPRegressor(hidden_layer_sizes=(300,300,50), activation='tanh') ֿךפ׿ָ㣐ֹ׉ֲ ת֮׉ֲַ׮ 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 x Ⰵ⸂ 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 y ⳿⸂ ✮庠　桦ָ׭׍ׯ ⡚ְז˘

Slide 161

Slide 161 text

/ 166 ֿךהֹ堣唒㷕统ח⡦ָ䗳銲ַ 111 ֿך邌ر٦ةד堣唒㷕统ٌرٕ׾鎮箺׃✮庠ׅ׸ל葺ְ 堣唒㷕统ٌرٕ ˖ ⯋稆穈䧭嫰 ˖ 㹋꿀勴⟝ ˖ 　桦 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 x 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 y Ⰵ⸂ ⳿⸂ MLPRegressor(hidden_layer_sizes=(300,300,50), activation='relu') ֲ٦׿ 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 x Ⰵ⸂ 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 y ⳿⸂ BDUJWBUJPO׾չ3F-6պח 頾⦼׾ئٗ縧䳔 ׻׶ה״ׁ׉ֲַ׮

Slide 162

Slide 162 text

/ 166 ֿךהֹ堣唒㷕统ח⡦ָ䗳銲ַ 112 ֿך邌ر٦ةד堣唒㷕统ٌرٕ׾鎮箺׃✮庠ׅ׸ל葺ְ 堣唒㷕统ٌرٕ ˖ ⯋稆穈䧭嫰 ˖ 㹋꿀勴⟝ ˖ 　桦 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 x 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 y Ⰵ⸂ ⳿⸂ KernelRidge(kernel='rbf', gamma=100.0, alpha=0.05) ֲ٦׿ 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 x Ⰵ⸂ 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 y ⳿⸂ 暴חך皘䨽כ ֿ׸ד葺ְךַ ،٦ذ؍ؿ؋ؙز

Slide 163

Slide 163 text

/ 166 ֿךהֹ堣唒㷕统ח⡦ָ䗳銲ַ 113 ֿך邌ر٦ةד堣唒㷕统ٌرٕ׾鎮箺׃✮庠ׅ׸ל葺ְ 堣唒㷕统ٌرٕ ˖ ⯋稆穈䧭嫰 ˖ 㹋꿀勴⟝ ˖ 　桦 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 x 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 y Ⰵ⸂ ⳿⸂ KernelRidge(kernel='laplacian', gamma=10.0, alpha=0.01) ֲ٦׿ AAACiXichVFNLwNRFD0dX1VfxUZi02iIVXNHBOmq0Y1lP7QkiMyMh9H5ysy0UU3/gJWdYEViIX6AH2DjD1j0J4gliY2F2+kkguBO3rzzzrvnvvPeVR1D93yiVkTq6u7p7Yv2xwYGh4ZH4qNjZc+uupooabZhu+uq4glDt0TJ131DrDuuUEzVEGtqJdveX6sJ19Nta9WvO2LLVPYsfVfXFJ+p8qZqNg6b2/EkpSiIxE8ghyCJMHJ2/A6b2IENDVWYELDgMzagwONvAzIIDnNbaDDnMtKDfYEmYqytcpbgDIXZCv/3eLURshav2zW9QK3xKQYPl5UJTNMj3dALPdAtPdH7r7UaQY22lzrPakcrnO2R44ni278qk2cf+5+qPz372MVS4FVn707AtG+hdfS1o9OXYrow3ZihK3pm/5fUonu+gVV71a7zonDxhx+VvfCLcYPk7+34CcpzKXkhRfn5ZGY5bFUUk5jCLPdjERmsIIcS1z/ACc5wLg1IsrQkpTupUiTUjONLSNkPIVKSSQ== x Ⰵ⸂ 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 y ⳿⸂

Slide 164

Slide 164 text

/ 166 ֿךהֹ堣唒㷕统ח⡦ָ䗳銲ַ 114 ֿך邌ر٦ةד堣唒㷕统ٌرٕ׾鎮箺׃✮庠ׅ׸ל葺ְ 堣唒㷕统ٌرٕ ˖ ⯋稆穈䧭嫰 ˖ 㹋꿀勴⟝ ˖ 　桦 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 x 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 y Ⰵ⸂ ⳿⸂ SVR(kernel='rbf', gamma=10, C=20) ֲ٦׿ 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 x Ⰵ⸂ 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 y ⳿⸂ 暴חך皘䨽כ ֿ׸ד葺ְךַ ،٦ذ؍ؿ؋ؙز

Slide 165

Slide 165 text

/ 166 ֿךהֹ堣唒㷕统ח⡦ָ䗳銲ַ 115 ֿך邌ر٦ةד堣唒㷕统ٌرٕ׾鎮箺׃✮庠ׅ׸ל葺ְ 堣唒㷕统ٌرٕ ˖ ⯋稆穈䧭嫰 ˖ 㹋꿀勴⟝ ˖ 　桦 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 x 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 y Ⰵ⸂ ⳿⸂ GaussianProcessRegressor(kernel=Matern(length_scale=100.0, nu=0.2)) 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 x Ⰵ⸂ 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 y ⳿⸂ 暴חך皘䨽כ ֿ׸ד葺ְךַ ،٦ذ؍ؿ؋ؙز ֲ٦׿

Slide 166

Slide 166 text

/ 166 ֿךהֹ堣唒㷕统ח⡦ָ䗳銲ַ 116 ֿך邌ر٦ةד堣唒㷕统ٌرٕ׾鎮箺׃✮庠ׅ׸ל葺ְ 堣唒㷕统ٌرٕ ˖ ⯋稆穈䧭嫰 ˖ 㹋꿀勴⟝ ˖ 　桦 AAACiXichVFNLwNRFD0dX1VfxUZi02iIVXNHBOmq0Y1lP7QkiMyMh9H5ysy0UU3/gJWdYEViIX6AH2DjD1j0J4gliY2F2+kkguBO3rzzzrvnvvPeVR1D93yiVkTq6u7p7Yv2xwYGh4ZH4qNjZc+uupooabZhu+uq4glDt0TJ131DrDuuUEzVEGtqJdveX6sJ19Nta9WvO2LLVPYsfVfXFJ+p8qZqNg6b2/EkpSiIxE8ghyCJMHJ2/A6b2IENDVWYELDgMzagwONvAzIIDnNbaDDnMtKDfYEmYqytcpbgDIXZCv/3eLURshav2zW9QK3xKQYPl5UJTNMj3dALPdAtPdH7r7UaQY22lzrPakcrnO2R44ni278qk2cf+5+qPz372MVS4FVn707AtG+hdfS1o9OXYrow3ZihK3pm/5fUonu+gVV71a7zonDxhx+VvfCLcYPk7+34CcpzKXkhRfn5ZGY5bFUUk5jCLPdjERmsIIcS1z/ACc5wLg1IsrQkpTupUiTUjONLSNkPIVKSSQ== x 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 y Ⰵ⸂ ⳿⸂ RandomForestRegressor(max_features='sqrt') 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 x Ⰵ⸂ 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 y ⳿⸂ ֲ٦׿

Slide 167

Slide 167 text

/ 166 ֿךהֹ堣唒㷕统ח⡦ָ䗳銲ַ 117 ֿך邌ر٦ةד堣唒㷕统ٌرٕ׾鎮箺׃✮庠ׅ׸ל葺ְ 堣唒㷕统ٌرٕ ˖ ⯋稆穈䧭嫰 ˖ 㹋꿀勴⟝ ˖ 　桦 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 x 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 y Ⰵ⸂ ⳿⸂ ExtraTreesRegressor() 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 x Ⰵ⸂ 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 y ⳿⸂ ֲ٦׿

Slide 168

Slide 168 text

/ 166 ֿךהֹ堣唒㷕统ח⡦ָ䗳銲ַ 118 ֿך邌ر٦ةד堣唒㷕统ٌرٕ׾鎮箺׃✮庠ׅ׸ל葺ְ 堣唒㷕统ٌرٕ ˖ ⯋稆穈䧭嫰 ˖ 㹋꿀勴⟝ ˖ 　桦 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 x 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 y Ⰵ⸂ ⳿⸂ ExtraTreesRegressor(bootstrap=True) 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 x Ⰵ⸂ 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 y ⳿⸂ TLMFBSOךرؿٕؓز כPזךד岣䠐 ֲ٦׿

Slide 169

Slide 169 text

/ 166 ֿךהֹ堣唒㷕统ח⡦ָ䗳銲ַ 119 ֿך邌ر٦ةד堣唒㷕统ٌرٕ׾鎮箺׃✮庠ׅ׸ל葺ְ 堣唒㷕统ٌرٕ ˖ ⯋稆穈䧭嫰 ˖ 㹋꿀勴⟝ ˖ 　桦 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 x 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 y Ⰵ⸂ ⳿⸂ GradientBoostingRegressor() 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 x Ⰵ⸂ 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 y ⳿⸂ ֲ٦׿

Slide 170

Slide 170 text

/ 166 ֿך⢽כ㹋כչ溪ךٌرٕظ؎ؤպך➂䊨ر٦ة 120 植㹋ך㹋庠ر٦ة ת׃גװ㜠デر٦ة חכ葿ղז衅ה׃瑎ָ

Slide 171

Slide 171 text

/ 166 ֿך⢽כ㹋כչ溪ךٌرٕظ؎ؤպך➂䊨ر٦ة 120 鵚ְ姻鍑⢽ָ JODPOTJTUFOU 侄䌌ظ؎ؤ 植㹋ך㹋庠ر٦ة ת׃גװ㜠デر٦ة חכ葿ղז衅ה׃瑎ָ

Slide 172

Slide 172 text

/ 166 ֿך⢽כ㹋כչ溪ךٌرٕظ؎ؤպך➂䊨ر٦ة 120 鵚ְ姻鍑⢽ָ JODPOTJTUFOU 侄䌌ظ؎ؤ Yך㼰׃ך麩ְד Zח䚈䃓ז㢌⻉ $MJT 植㹋ך㹋庠ر٦ة ת׃גװ㜠デر٦ة חכ葿ղז衅ה׃瑎ָ

Slide 173

Slide 173 text

/ 166 ֿך⢽כ㹋כչ溪ךٌرٕظ؎ؤպך➂䊨ر٦ة 120 鵚ְ姻鍑⢽ָ JODPOTJTUFOU 侄䌌ظ؎ؤ Yך㼰׃ך麩ְד Zח䚈䃓ז㢌⻉ $MJT 㢩׸⦼ 植㹋ך㹋庠ر٦ة ת׃גװ㜠デر٦ة חכ葿ղז衅ה׃瑎ָ

Slide 174

Slide 174 text

/ 166 ֿך⢽כ㹋כչ溪ךٌرٕظ؎ؤպך➂䊨ر٦ة 120 ر٦ةָזְ㼰זְ걄㚖 鵚ְ姻鍑⢽ָ JODPOTJTUFOU 侄䌌ظ؎ؤ Yך㼰׃ך麩ְד Zח䚈䃓ז㢌⻉ $MJT 㢩׸⦼ 植㹋ך㹋庠ر٦ة ת׃גװ㜠デر٦ة חכ葿ղז衅ה׃瑎ָ

Slide 175

Slide 175 text

/ 166 ֿך⢽כ㹋כչ溪ךٌرٕظ؎ؤպך➂䊨ر٦ة 121 Neural Networks (ReLU) Random Forest Extra Trees (bootstrap) Neural Networks (Tanh) Linear Regression Kernel Ridge (RBF) Kernel Ridge (Laplacian) Extra Trees (no bootstrap) Gradient Boosting “ਅͷ”Ϟσϧ

Slide 176

Slide 176 text

/ 166 3BTIPNPO⸬卓♧⡤ו׸׾⥋ׄ׸לְְ׿ׄׯְ 122 $堣唒㷕统ٌرٕװ鎮箺ر٦ةָ㢌׻׸ל✮庠כ㢌׻׷ ٌرٕהر٦ةך侧׌ֽ✮庠׮֮׷溪㹋כչ谞ך⚥պ ˖ $SPTTWBMJEBUJPO礵䏝כקרず瘝ךٌرָٕ搀侧ח֮׶䖤׷ ˖ ずٌׄرٕד׮)ZQFSQBSBNFUFSָ麩ִלַז׶麩ְ䖤׷ ˖ 植㹋דכ溪ךٌرٕכⴓַ׵זְ׋׭葺׃䝤׃ךⴻ倖כ㔭ꨇ ˖ Ⰵ⸂㢌侧׾䙼ְאֻ׌ֽⰅ׸ג넝如⯋ر٦ةחז׷הׁ׵ח 3BTIPNPO⸬卓ٔأָؙ㟓㣐 猘ך如⯋⢽ך⡲捀䚍ח׀岣䠐 .-ָוְֲֲ䪮遭זךַ.-ך暴䚍הꣲ歲׾姻׃ֻ䪾䳢ׅ׷ ✮庠穠卓׾䔲鑩ⴓꅿך濼陎ח撑׵׃ג岣䠐帾ֻ嗚鏾٥鍑ꅸׅ׷

Slide 177

Slide 177 text

/ 166 ׍ז׫ח؟ٝفٕ侧ָ⼧ⴓ㣐ֹֽ׸ל׻׶הו׸ד׮0,123 ؟ٝفٕ侧ָ⼧ⴓ㢳ֽ׸ל⢽㢩涸ر٦ة挿כ窟鎘涸ח湱媷ׁ׸׷ ̔׋׌׃넝如⯋ד֮׷קו䭷侧涸ז侧ָ䗳銲דꬊ植㹋涸ז劍䖉 Neural Networks (ReLU) Random Forest Extra Trees (bootstrap)

Slide 178

Slide 178 text

/ 166 㢳ֻכ䙀걧ח֮׷⦪酡瑞꟦ח㼎׃ر٦ةָ駈׶גְזְ 124 椚䟝涸זر٦ة ࣮ࡍʹखʹೖΔσʔλ (Underspecification) Neural Networks (ReLU) Random Forest Extra Trees (bootstrap) Neural Networks (ReLU) Random Forest Extra Trees (bootstrap) Neural Networks (Tanh) Linear Regression Kernel Ridge (RBF) Kernel Ridge (Laplacian) Extra Trees (no bootstrap) Gradient Boosting

Slide 179

Slide 179 text

/ 166 㢳ֻכ䙀걧ח֮׷⦪酡瑞꟦ח㼎׃ر٦ةָ駈׶גְזְ 124 椚䟝涸זر٦ة ࣮ࡍʹखʹೖΔσʔλ (Underspecification) Neural Networks (ReLU) Random Forest Extra Trees (bootstrap) Neural Networks (ReLU) Random Forest Extra Trees (bootstrap) Neural Networks (Tanh) Linear Regression Kernel Ridge (RBF) Kernel Ridge (Laplacian) Extra Trees (no bootstrap) Gradient Boosting ໘౗ͳ͜ͱʹ͜͜ͰRashomon͕ى͖ͯ͠·͏ʂ ͜ͷ΁ΜͷϞσϧ͸खʹೖΔࣄྫͰͷCVਫ਼౓͸ ಉϨϕϧ͕ͩςετͰͷڍಈ͸͍ͩͿҟͳΔʂ

Slide 180

Slide 180 text

/ 166 -FP#SFJNBOךٖحأٝ 125 ➙ד׮ֿך挿כת׌葿ղה0QFOז㉏겗׾㶱׿ד灇瑔ׁ׸גְ׷ Ӎ "Rashomon"葺ְٌرٕך㢳ꅾ䚍 ꬊ♧䠐䚍 ず玎䏝ך葺ְ✮庠礵䏝׾䭯אⰋֻ殯ז׷ٌرָٕ׋ֻׁ׿㶷㖈ׅ׷ Ӎ "Occam"ٌرٕך鍑ꅸ䚍ה✮庠礵䏝ך؝ٝؿؙٔز ٌرٕךءٝفׁٕ 鍑ꅸ䚍 ה✮庠礵䏝ך⚕甧כהג׮ꨇ׃ְ Ӎ "Bellman"넝如⯋ر٦ةָ䒷ֹ饯ֿًׅٔحزהرًٔحز 넝如⯋ז邌植 ꟼ⤘׃׉ֲזדֹ׷׌ֽ㢳ֻך㢌ꆀ ׾䪔ֲץֹזךַ ⠗窟涸ז窟鎘㷕ך״ֲח佄ꂁ涸ז㼰侧ך㢌ꆀ׾嗚鎢׃ⴓ匿ׅץֹזךַ ̔ؗحثٝءؙٝ㔐䌓 䙼ְאֻ㢌侧Ⰻ鿇Ⰵ׶ٌرٕ ٥暴䗙ꆀ ؒٝآٔ،ؚٔٝה亻⡂湱ꟼ٥3BTIPNPO⸬卓ך㟓㣐ٔأؙ 3BTIPNPO 0DDBN #FMMNBO

Slide 181

Slide 181 text

/ 166 堣唒㷕统ך✮庠⦼כ鎮箺ر٦ةך剑葺⦼׾סאֲ馄ִזְ 126 堣唒㷕统ٌرٕכ劍䖉铎䊴ָ剑㼭חז׷״ֲ 鎮箺ر٦ةך溪׿⚥׾鸐׷״ֲ חؿ؍حذ؍ׁؚٝ׸׷׋׭ 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 x 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 x1 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 x2 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 y ر٦ةך剑㣐⦼ ر٦ةך剑㼭⦼ 堣唒㷕统ך ✮庠⦼ סאֲֿך眔㔲

Slide 182

Slide 182 text

/ 166 堣唒㷕统ך✮庠⦼כ鎮箺ر٦ةך剑葺⦼׾סאֲ馄ִזְ 126 堣唒㷕统ٌرٕכ劍䖉铎䊴ָ剑㼭חז׷״ֲ 鎮箺ر٦ةך溪׿⚥׾鸐׷״ֲ חؿ؍حذ؍ׁؚٝ׸׷׋׭ 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 x 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 x1 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 x2 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 y ̔傀濼ך׮ך״׶葺ְ׮ך׾鋅אֽ׷הְֲ䱱稊湡涸ח♶黝さ ر٦ةך剑㣐⦼ ر٦ةך剑㼭⦼ 堣唒㷕统ך ✮庠⦼ סאֲֿך眔㔲

Slide 183

Slide 183 text

/ 166 堣唒㷕统ך✮庠⦼כ鎮箺ر٦ةך剑葺⦼׾סאֲ馄ִזְ 127 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 x ׁ׵חծ׮׃✮庠⦼ָ鎮箺ر٦ةך剑葺⦼׾♳㔐׷ה׃ג׮ծ ر٦ةָזְ걄㚖ד׉ך✮庠כ⟣䠐涸ד䔲גחדֹזְ˘ ر٦ةך 剑㣐⦼ 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 y

Slide 184

Slide 184 text

/ 166 堣唒㷕统ך✮庠⦼כ鎮箺ر٦ةך剑葺⦼׾סאֲ馄ִזְ 128 SVR(kernel='rbf', gamma=10, C=100) 如⯋ך⢽דכչⰻ䯏պַչ㢩䯏պַⴓַ׷䠬ָׄ׃ג׃תֲָ ♧菙ך넝如⯋דכֿךⴻⴽׅ׵湫䠬涸דכזְֿהח岣䠐 ֿֿכⰻ䯏㢩䯏ر٦ة㢩걄㚖

Slide 185

Slide 185 text

/ 166 չر٦ة㢩걄㚖דך䠐㔳׃זְ㢩䯏պٔأؙכ䩛岀ח⣛㶷 129 寸㹀加،ٝ؟ٝـٕ岀 3BOEPN'PSFTU瘝 דכֿך朐屣כ⾱椚♳ 饯ֿ׵זְָ简䕎㔐䌓װ/FVSBM/FUXPSLTזו➭ך䩛岀דכ岣䠐 Linear Regression Kernel Ridge (RBF) Gradient Boosting Neural Networks (Tanh) Kernel Ridge (Laplacian) Extra Trees (no bootstrap) Neural Networks (ReLU) Random Forest Extra Trees (bootstrap)

Slide 186

Slide 186 text

/ 166 ✮庠⦼׌ֽדכזֻ׉ך✮庠ⴓ侔 然䏝 ׮罋ִ׷ךָꅾ銲 130 AAACiXichVFNLwNRFD0dX1VfxUZi02iIVXNHBOmq0Y1lP7QkiMyMh9H5ysy0UU3/gJWdYEViIX6AH2DjD1j0J4gliY2F2+kkguBO3rzzzrvnvvPeVR1D93yiVkTq6u7p7Yv2xwYGh4ZH4qNjZc+uupooabZhu+uq4glDt0TJ131DrDuuUEzVEGtqJdveX6sJ19Nta9WvO2LLVPYsfVfXFJ+p8qZqNg6b2/EkpSiIxE8ghyCJMHJ2/A6b2IENDVWYELDgMzagwONvAzIIDnNbaDDnMtKDfYEmYqytcpbgDIXZCv/3eLURshav2zW9QK3xKQYPl5UJTNMj3dALPdAtPdH7r7UaQY22lzrPakcrnO2R44ni278qk2cf+5+qPz372MVS4FVn707AtG+hdfS1o9OXYrow3ZihK3pm/5fUonu+gVV71a7zonDxhx+VvfCLcYPk7+34CcpzKXkhRfn5ZGY5bFUUk5jCLPdjERmsIIcS1z/ACc5wLg1IsrQkpTupUiTUjONLSNkPIVKSSQ== x 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 x1 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 x2 䱱稊חꟼׅ׷䠐䙼寸㹀ח崞欽ׅ׷ךד֮׸ל堣唒㷕统ٌرٕך ✮庠⦼ךⴓ侔ⴓ䋒⥋걾⼒꟦׾罋ִ׷ֿהָꅾ銲 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 y ر٦ةך剑㣐⦼ FHչ　桦✮庠⦼כpպWTչ　桦✮庠⦼כpպ

Slide 187

Slide 187 text

/ 166 䱱稊ׅ׷ꥷכ✮庠⦼荈⡤׾䭷垥ה׃זְ 131 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 x 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 x1 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 x2 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 y 䱱稊ך湡涸דכ劍䖉何㊣ &* װ⥋걾⼒꟦ך♳ꣲזו׾䭷垥ח ر٦ةך剑㣐⦼ 劍䖉何㊣ &* 　桦何㊣ꆀך劍䖉⦼ 　桦何㊣ꆀ׾剑㣐⦼׾馄ִ׷鿇ⴓך然桦ךꅾ׫ד琎ⴓ

Slide 188

Slide 188 text

/ 166 俑柃ַ׵꧊׭׋㹋ꥷך㹋꿀ر٦ة㜠デ׾⢪ֲ 132 http://www.fhi-berlin.mpg.de/acnew/department/pages/ocmdata.html https://www.nature.com/articles/s41467-019-08325-8#Sec19 Oxidative coupling of methane (OCM) reactions Methane (CH4 ) is partially oxidized to C2 hydrocarbons such as ethane (C2 H6 ) and ethylene (C2 H4 ) in a single step Elemental composition of catalyst (mol%) Process parameters + Preparation Catalytic performance • Zavyalova, U.; Holena, M.; Schlögl, R.; Baerns, ChemCatChem 2011. • Followup: Kondratenko, E. V.; Schlüter, M.; Baerns, M.; Linke, D.; Holena, M. Catal. Sci. Technol. 2015. • Renalysis with Corrections & Outlier Removal Schmack, R.; Friedrich, A.; Kondratenko, E. V.; Polte, J.; Werwatz, A.; Kraehnert, R. Nat Commun 2019. 1866 catalyst records from 421 reports

Slide 189

Slide 189 text

/ 166 俑柃ַ׵　꧊׃׋ر٦ةך㉏겗 133 Ԩ 6OEFSTQFDJDBUJPO㹋꿀勴⟝ָ䓼ֻ䕦갟ׅ׷ָծ醱侧ך㹋꿀勴⟝ד 㜠デ⢽ך֮׷鍗㯭כ噰׭ג㼰זְկず♧勴⟝דך醱醡㹋꿀׮ז׃կ ̔⢽⚥ծ勴⟝⟃♳⢽ծ勴⟝⟃♳⢽ծ勴⟝⟃♳⢽ Ԩ 4QBSTJUZ穈䧭ד⢪׻׸׷⯋稆ךؔ٦غٓحفָ㼰זֻꬊ䌢חأػ٦أ ̔⢽ִל/B5J.Oˏה;O$Fˏ׾וֲ嫰鯰ׅץֹ 74 elements All pairwise comparisons

Slide 190

Slide 190 text

/ 166 俑柃ַ׵　꧊׃׋ر٦ةך㉏겗 134 Ԩ 4USPOH#JBT⳿晛ׁ׸׷猰㷕䧭卓ך䓼ְ䧭⸆غ؎،أծ崧遤װ㹋꿀ך ׃װׁׅח״׷鼅䫛غ؎،أծזו钠濼غ؎،أ٥爡⠓涸غ؎،أך䕦갟 LaO3, Li/MgO, Mn/Na2WO4/SiO2 זו 暴㹀ך鍗㯭ָꬊ䌢ח״ֻ灇瑔ׁ׸׷

Slide 191

Slide 191 text

/ 166 堣唒㷕统✮庠ך然䏝٥⥋걾䏝׮鋅ג黝欽眔㔲׾䢅ꅾח椚鍑 135 Gradient Boosted Trees Extra Trees (no bootstrap) Random Forest Extra Trees (bootstrap) sample max sample min GradientBoostingRegressor LGBMRegressor RandomForestRegressor ExtraTreesRegressor By quantile regression to .16, .5, .84 quantiles Naturally by the law of total variance bounded prediction Ԩ 寸㹀加،ٝ؟ٝـٕך✮庠ⴓ侔 ⥋걾⼒꟦ 鎮箺ر٦ةךNBYNJOך꟦ח✮庠⦼ָֻ ׷ךד䠐㔳׃זְ㢩䯏ٔأָؙ㼰זְ

Slide 192

Slide 192 text

/ 166 暴䗙ꆀך鏣鎘鍗㯭ך⸬卓涸ז暴䗙邌植׾罋ִ׷ 136 ˖ ֿךתתדכծ涯瀧祩։ֹן瀧祩תדכءّ祩דչ欥㄂պծꄣ屘ח כչ㞁ⴓպָろת׸׷ծזוծ銲稆ךչ⦐䚍պכⰋֻ罋䣁ׁ׸זְկ ˖ 鎮箺ر٦ةחろת׸זְ銲稆ָⰅ׷ה✮庠ח⿾僥דֹזְկ ˖ 銲稆׀הך걼䏝ָكؗ⛦⵱涸ד֮׶ꬊ䌢ח㣐ֹז⨉׶ָ֮׷կ ˖ 銲稆侧ָ㢳ֻ㜠デ⢽ח銲稆ךؔ٦غٓحفָ㼰זְկ 涯瀧祩 ♲庛祩 ؚٓصُ٦祩 랲祩 ג׿ְׁ祩 ֹן瀧祩 㞁 ꄣ屘 ֲת׫锃㄂俱 ֿ׃׳ֲ ءشٌٝ ؖ٦ٔحؙ ؖٓيو؟ٓ ٕؕتٌٝ ؙىٝ ؟ؿٓٝ ؝ٔ،ٝت٦ ة٦ًٔحؙ شخًؚ ׻ׁן ׃׳ֲָ ְֶ׃ׁ 湡涸㢌侧 锃椚勴⟝ 銲稆穈䧭

Slide 193

Slide 193 text

/ 166 ⯋稆ך⦐䚍׾⯋稆鎸鶢㶨كؙزٕד邌ׅⰅ⸂邌植 137 ⯋稆׾չءٝنٕպה׃ג䪔ֲךדכזֻ չ㢳如⯋ך⯋稆鎸鶢㶨كؙزٕպד䪔ֲ ⯋稆鎸鶢㶨ך䬄韋䏝׾㢌ִ׸לծꟼ䗰ך֮׷ 暴䚍ך׫ח滠湡׃ג⯋稆ך邌植٥嫰鯰ָ〳腉ח 鎮箺ر٦ةחろת׸זְ⯋稆׮䪔ִ׷

Slide 194

Slide 194 text

/ 166 ⯋稆ך⦐䚍׾⯋稆鎸鶢㶨كؙزٕד邌ׅⰅ⸂邌植 138 48&% 4PSUFE8FJHIUFE&MFNFOUBM%FTDSJQUPST 邌植 穈䧭嫰–⯋稆鎸鶢㶨كؙزٕ׾穈䧭嫰ך꣬갫ח⚛ץ׋׮ך ̔ءٝفٕ׌ָ㹀ꆀ涸ז何㊣ָ䖤׵׸׋暴䗙كؙزٕ邌植 +

Slide 195

Slide 195 text

/ 166 48&%邌植ה欽ְ׷⯋稆鎸鶢㶨ח״׷䬄韋䏝ךⵖ䖴 139 74 elements Compositional (onehot-like) Catalyst: Mg 83.46, Li 16.53 SWED-8 83.46 × 16.53 × 0.00 0.00 … SWED-3 83.46 × 16.53 × 0.00 0.00 … SWED-3 features: electronegativity, density, enthalpy of fusion SWED-8 features: SWED-3 features + atomic weight, atomic radius, m.p., b.p., ionization enegy can control specificity & focus Ԩ ⯋稆鎸鶢㶨ה׃גוך״ֲז׮ך׾⢪ֲַח״׏ג磦鋔⻉ָⵖ䖴〳腉 Ԩ ⯋稆כ鼅׿׌⯋稆鎸鶢㶨ך侧⦼ד邌植ׁ׸ծֿך邌植ך׮הדⰻ䯏ׁ׸׷ ׋׭ծ鎮箺ر٦ةחכזְ⯋稆׮荈搫ח《׶䪔ֲֿהָדֹ׷կ

Slide 196

Slide 196 text

/ 166 倜׃ְزٖٝس׾⿾僥ׅ׷׋׭ر٦ةإحز荈⡤׮䭁⯍ 140 The original dataset: 1866 catalyst records from 421 reports (1982 - 2009) Mine, S.; Takao, M.; Yamaguchi, T.; Toyao, T.*; Maeno, Z.; Hakim Siddiki, S. M. A.; Takakusagi, S.; Shimizu, K.*; Takigawa, I.* ChemCatChem 2021. https://doi.org/10.1002/cctc.202100495. 4559 catalyst records from 542 reports The update dataset: 4559 catalyst records from 542 reports (2010 - 2019)

Slide 197

Slide 197 text

/ 166 $　桦ך堣唒㷕统✮庠 141 1. Conventional: composition + condition 2. Proposed(Exploitative): composition + SWED + condition 3. Proposed(Explorative): SWED + condition w/ SWED→composition estimator RFR (Random Forest); ETR (ExtraTrees); XGB (XGBoost) SWED-3 features: electronegativity, density, enthalpy of fusion SWED-8 features: SWED-3 features + atomic weight, atomic radius, m.p., b.p., ionization enegy

Slide 198

Slide 198 text

/ 166 48&%׾欽ְ׋鍗㯭⦪酡 劍䖉何㊣⦼ד♳⡘⦐ 142 鎮箺ر٦ةחזְ⯋稆 "T )G 4F 0T 1N ׮䲿 周⦪酡ח鋅׵׸׋ 嫩䚍ך׋׭㹋欽♳כ㉏ 겗ָ֮׷׮ךךծ׉ך ״ֲז⦪酡׮ֹ׍׿ה 䱱稊ׁ׸גְ׷

Slide 199

Slide 199 text

/ 166 㷕统חכչ濼陎ךⵃ欽պהչ䱱稊պךزٖ٦سؔؿָ⠵ֲ 143 倜׃ְֿה׾չ㷕עպꥷך剑׮㛇劤涸זزٖ٦سؔؿ ➙תד㷕׿׌ֿהךչⵃ欽պ ➙תדךر٦ةך堣唒㷕统ח㛇בֻ✮庠ך崞欽 ➙תדח㷕׿דזְֿהךչ䱱稊պ 倜׃ְ穗꿀ծ倜׃ְ濼陎ךェ　ծ濼陎ך䭁⯍ ˟Ⰻ⡤ח⽑׭׷չ➙תדח㷕׿׌ֿהպךؕغ٦桦ָ ⡚ְ㜥さכ加׾鋅ג啾׾鋅׆חז׏ג׃תֲ ➙תדךر٦ةך然䏝ָ⡚ְ걄㚖ծر٦ةָזְ걄㚖ַ׵ך 刿ז׷ر٦ةך《䖤

Slide 200

Slide 200 text

/ 166 俑柃ر٦ةחכ圫ղז㉏겗ָ֮׶չ䱱稊պָ״׶ꅾ銲 144 Ԩ 㹋꿀כ➂꟦ָ鎘歗ׅ׷׋׭ծ钠濼غ؎،أװ爡⠓涸غ؎،أָ⿾僥ׁ׸ ג׃תֲ 䖞勻濼鋅ծ崧遤ծ⠗窟ծ㹋꿀׃װׁׅ˘ Ԩ 鎮箺ر٦ةך✲⢽ךⴓ䋒ח㣐ֹז⨉׶ָ֮׷ָծֿ׸כ荈搫ך䶏椚דכ זֻ猘׋׍ך鋔ꅿך杞ׁ 䙼ְ鴥׫ ׾⿾僥׃׋׮ך Ԩ وة؎⸬卓 .BUUIFXFFDU 䧭⸆⢽ח麓ⶱח䒷ֹ׆׵׸ָ׍ Ԩ 䧭⸆⢽ך׫ָ㜠デׁ׸׷׋׭㣟侁✲⢽ך䞔㜠ָ荜ㄏ涸ח妀䴦

Slide 201

Slide 201 text

/ 166 㹋꿀鎘歗ה堣唒㷕统 145 ˖ 䱱稊挿׾鎘歗דֹ׷㜥さכծ欰饯䟝㹀眔㔲חדֹ׷׌ֽչת׿ץ׿ז ֻպה׷קֲָ葺ְխ FHٓٝتي㹋꿀ծ㸣Ⰻ㹋倵銲㔓鎘歗ծٓذٝ 馄倯呓鎘歗ծ%剑黝鎘歗ծ˘ ˖ 䱱稊װ㹋꿀鎘歗חְֶגכ.-כ植㹋ך➿椚ٌرٕחֺׅזְ ֿך✲⢽ⴓ䋒כ 䱱稊חכ♶銲 ずׄ⢽侧ד嚊䕎ָ ⴓַ׶.-׮⸬卓涸

Slide 202

Slide 202 text

/ 166 㹋꿀鎘歗חֶֽ׷ؿ؍حءٍ٦ך♲⾱⵱ 146

Slide 203

Slide 203 text

/ 166 㹋꿀鎘歗חֶֽ׷ؿ؍حءٍ٦ך♲⾱⵱ 146 ⿾䗁SFQMJDBUJPO ̔ず勴⟝ד醱侧㔐ך㹋꿀׾遤ֲկⱄ植䚍ך䬐⥂ח⸇ִծ խֿך䞔㜠ָזְה禸窟铎䊴ה⩐搫铎䊴׾ⴻⴽדֹזְկ

Slide 204

Slide 204 text

/ 166 㹋꿀鎘歗חֶֽ׷ؿ؍حءٍ٦ך♲⾱⵱ 146 ⿾䗁SFQMJDBUJPO ̔ず勴⟝ד醱侧㔐ך㹋꿀׾遤ֲկⱄ植䚍ך䬐⥂ח⸇ִծ խֿך䞔㜠ָזְה禸窟铎䊴ה⩐搫铎䊴׾ⴻⴽדֹזְկ 搀⡲捀⻉SBOEPNJ[BUJPO ̔罋ִ׋ְ銲㔓⟃㢩ח湡涸㢌侧ח䕦갟׾♷ִ׷〳腉䚍ָ֮׷銲㔓ָ ֮׷㜥さծ〳腉זꣲ׶ٓٝتيחⶴ׶➰ֽׅ׷կ DG穠卓ָ״ַ׏׋勴⟝ת׻׶ד㢳׭ח鑐׃׋ְךכ椚鍑דֹ׷ ָ䱱稊ָ湡涸ז׵׬׃׹ٓٝتي㹋꿀勴⟝ךקֲָ葺ְ

Slide 205

Slide 205 text

/ 166 㹋꿀鎘歗חֶֽ׷ؿ؍حءٍ٦ך♲⾱⵱ 146 ⿾䗁SFQMJDBUJPO ̔ず勴⟝ד醱侧㔐ך㹋꿀׾遤ֲկⱄ植䚍ך䬐⥂ח⸇ִծ խֿך䞔㜠ָזְה禸窟铎䊴ה⩐搫铎䊴׾ⴻⴽדֹזְկ 搀⡲捀⻉SBOEPNJ[BUJPO ̔罋ִ׋ְ銲㔓⟃㢩ח湡涸㢌侧ח䕦갟׾♷ִ׷〳腉䚍ָ֮׷銲㔓ָ ֮׷㜥さծ〳腉זꣲ׶ٓٝتيחⶴ׶➰ֽׅ׷կ DG穠卓ָ״ַ׏׋勴⟝ת׻׶ד㢳׭ח鑐׃׋ְךכ椚鍑דֹ׷ ָ䱱稊ָ湡涸ז׵׬׃׹ٓٝتي㹋꿀勴⟝ךקֲָ葺ְ 㽷䨽盖椚MPDBMDPOUSPM ̔罋ִ׋ְ銲㔓⟃㢩ךغحؙؚٓؐٝس㔓㶨כדֹ׷׌ֽ խ㖱♧חז׷״ֲח㹋꿀׾盖椚ׅ׷կ DG㹋꿀勴⟝ך剑黝⻉כ镘׭ג㔿㹀׃穈䧭׌ֽס׷㹋꿀

Slide 206

Slide 206 text

/ 166 㹋꿀鎘歗ה堣唒㷕统 147 C2 yield Expected Improvement ML surrogate w/ 95%CI n given data points Input representation (SWED + Exp. Cond.) .-חⰅ⸂ׅ׷ر٦ةחזְ⫘ぢכ⾱椚♳✮庠דֹזְ׋׭ ،ٕ؞ٔؤيך鑫稢״׶׮ر٦ةך　꧊鎘歗 㹋꿀鎘歗 ծ 黝欽眔㔲ך椚鍑ծㅷ颵⥂鏾ָ䧭⸆ך꒲ד֮׷ֿה׾ְא׮䗰ח 葺ְ暴䗙邌植ה✮庠礵䏝 ך넝ְ➿椚ٌرٕ ✮庠ⴓ䋒׾罋䣁׃׋䱱稊 䭷垥ך㽷䨽ؾ٦ؙ̕ず㹀 如ח㹋꿀ׅ׷⣣⦼ך넝ְ ⦪酡挿׾EJWFSTJGZ׃䲿爙 ⡂׋⦪酡挿כؚٕ٦ف⻉׃➿邌挿׾邌爙

Slide 207

Slide 207 text

/ 166 堣唒㷕统ٌرָٕז׈׉ך✮庠׾׃׋ַך銲㔓ⴓ匿 148 堣唒㷕统ٌرٕ ˖ ⯋稆穈䧭嫰 ˖ 㹋꿀勴⟝ ˖ 　桦 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 x Ⰵ⸂ ⳿⸂ 　桦խָ넝ְ鍗㯭ה⡚ְ鍗㯭ך麩ְ׾鋉㹀׃ֲ׷ 㔓㶨כ⡦׌׹ֲ ׋׌׃Ⰵ⸂ָろ׬䞔㜠ך眔㔲ד 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 y AAACiXichVFNLwNRFD0dX1VfxUZi02iIVXNHBOmq0Y1lP7QkiMyMh9H5ysy0UU3/gJWdYEViIX6AH2DjD1j0J4gliY2F2+kkguBO3rzzzrvnvvPeVR1D93yiVkTq6u7p7Yv2xwYGh4ZH4qNjZc+uupooabZhu+uq4glDt0TJ131DrDuuUEzVEGtqJdveX6sJ19Nta9WvO2LLVPYsfVfXFJ+p8qZqNg6b2/EkpSiIxE8ghyCJMHJ2/A6b2IENDVWYELDgMzagwONvAzIIDnNbaDDnMtKDfYEmYqytcpbgDIXZCv/3eLURshav2zW9QK3xKQYPl5UJTNMj3dALPdAtPdH7r7UaQY22lzrPakcrnO2R44ni278qk2cf+5+qPz372MVS4FVn707AtG+hdfS1o9OXYrow3ZihK3pm/5fUonu+gVV71a7zonDxhx+VvfCLcYPk7+34CcpzKXkhRfn5ZGY5bFUUk5jCLPdjERmsIIcS1z/ACc5wLg1IsrQkpTupUiTUjONLSNkPIVKSSQ== x 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 x Ⰵ⸂ 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 y ⳿⸂ ر٦ةإحز 䎂㖱

Slide 208

Slide 208 text

/ 166 堣唒㷕统ٌرָٕז׈׉ך✮庠׾׃׋ַך銲㔓ⴓ匿 149 SHAP (SHapley Additive exPlanations) https://speakerdeck.com/dropout009/shapley-additive-explanationsdeji-jie-xue-xi-moderuwojie-shi-suru ♷ִ׵׸׋✮庠⦼ךر٦ةإحز䎂㖱ַ׵ך㢌⻉ꆀ׾ չ暴䗙ꆀ׀הך㺔♷䏝 4)"1⦼ ךㄤպפⴓ鍑ׅ׷ٌرٕ铡僇岀 و؎شأ銲㔓׮֮׶ֲ׷

Slide 209

Slide 209 text

/ 166 堣唒㷕统ٌرָٕז׈׉ך✮庠׾׃׋ַך銲㔓ⴓ匿 150 Composition: (1) Mg 83.46 (2) Li 16.53 ]4)"1⦼] ך꣬갫 4)"1⦼ךㄤر٦ةإحز䎂㖱ַ׵ך㟓⸇ⴓ 㢳➂侧ך⼿⸂؜٦يד䖤׋㜠ꂹ׾ぐفٖ؎َפⰕ䎂חⴓꂁׅ׷ ؜٦ي椚锷ך㉏겗ה׫זֿׅהדぐղך暴䗙ꆀך㺔♷䏝׾皾⳿

Slide 210

Slide 210 text

/ 166 5SFF&YQMBJOFS寸㹀加،ٝ؟ٝـٕ欽ך4)"1 151 ؎ٝةؙٓذ؍ـז鍑匿׾䲿⣘ׅ׷הג׮ֿז׸׋خ٦ٕ׮֮׷ ̔ ♧菙חכ鎘皾㔭ꨇ /1㔭ꨇ זꆀ׌ָծ寸㹀加،ٝ؟ٝـٕדכ 4)"1⦼ָ㢳갪䒭実鍑〳腉 5SFF&YQMBJOFSPSUSFF4)"1 https://github.com/slundberg/shap

Slide 211

Slide 211 text

/ 166 4)"1ח״׷㷕统幥׫ٌرַٕ׵ך銲㔓ⴓ匿 152 ر٦ةװ㷕统׃׋ٌرַٕ׵䖤׵׸׷㢳錬涸䞔㜠׾〳鋔⻉זו ד䬄⳿׃ծ㼔Ꟍ㹺ה⼿⫴׃㼔Ꟍ濼鋅װ㹋ⵖ秈ח撑׵׃גⵃ崞欽 Feature Importance Partial Dependence Interaction Effect

Slide 212

Slide 212 text

/ 166 4)"1ח״׷㷕统幥׫ٌرַٕ׵ך銲㔓ⴓ匿 153 1st: (1) Mn: 72.3 (2) Li: 27.7 2nd: (1) Sr:50.0 (2) Ce:45.0 (3) Yb:5.0 Mine, S.; Takao, M.; Yamaguchi, T.; Toyao, T.*; Maeno, Z.; Hakim Siddiki, S. M. A.; Takakusagi, S.; Shimizu, K.*; Takigawa, I.* ChemCatChem 2021. https://doi.org/10.1002/cctc.202100495.

Slide 213

Slide 213 text

/ 166 侄鎮 154 荈搫猰㷕ⴓꅿדךⵃ崞欽כ.-ך䪮遭灇熊׌ֽדכ䧭⸆׃זְկ ⴓꅿ㼔Ꟍ㹺הך⼿⫴ָ䗳銲♶〳妀 ˖ .-ָוְֲֲ䪮遭זךַ.-ך暴䚍הꣲ歲׾姻׃ֻ䪾䳢ׅ׷ ˖ չر٦ةך　꧊鎘歗 㹋꿀鎘歗 הㅷ颵⥂鏾ծ黝欽眔㔲ך椚鍑պ ָˑEBUBESJWFO˒ך䗰茖ד֮׷ֿה׾ְא׮䗰ח ˖ չ䱱稊պָ湡涸ז׵.-ך卓׋ׅ䕵ⶴכֻ֮תד♧鿇ה䗰䖤׷ " 㼔Ꟍ㹺הך⼿⫴ծⴓꅿך㼔Ꟍ濼陎ח撑׵׃׋嗚鏾٥鍑ꅸ " ءىُٖ٦ءّٝ٥㹋꿀荈⹛⻉٥锷椚䱿锷הך輐さ

Slide 214

Slide 214 text

/ 166 ➙傈ךذ٦و 155 ˖ 荈䊹稱➜ 堣唒㷕统ה荈搫猰㷕ך㞮歲 ˖ 堣唒㷕统הכ倜׃ְفؚٗٓىؚٝך倯岀 ˖ 堣唒㷕统㾊כ♧⡤⡦ָ嚂׃ְךַ ˖ ⴓ㶨ך邌植ה堣唒㷕统 ˖ ؚٖ؎نحؙأ剑黝⻉ 怴糊䌓秛 锷椚㷕ה窟鎘㷕ך輐さ ˖ 荈搫猰㷕灇瑔ד堣唒㷕统׾⢪ֲֶהׅ׷ה䗳׆עאַ׷劤䔲חꨇ׃ְ㉏겗 ˖ ر٦ةٌرؚٔٝה✮庠،ٕ؞ٔؤي 5IF5XP$VMUVSFT ˖ ✮庠ַ椚鍑ַ3BTIPNPO⸬卓 6OEFSTQFDJDBUJPO 鍑ꅸ㢳圫䚍 ˖ ➂꟦ך钠濼غ؎،أח歋勻ׅ׷㉏겗⟎铡ծ㣟侁ծ䧭⸆غ؎،أծFUD ˖ 堣唒㷕统ַ׵堣唒涪鋅פ ˖ չ涪鋅պչ椚鍑պך麣瘡כさ椚⻉דֹ׷ךַ荈⹛⻉דֹ׷ךַ

Slide 215

Slide 215 text

/ 166 堣唒涪鋅כ堣唒㷕统ך䪮遭灇熊׌ֽדכⵋ麦׃ִזְ 156 猰㷕ָ実׭׷ֿהⴓַ׵זַ׏׋ֿהָⴓַ׷ 猰㷕涸涪鋅ה猰㷕涸椚鍑 涪鋅 椚鍑 ⾱㔓ה穠卓 㔓卓ꟼ⤘ ׾鋅⳿ׅ ׉׃ג㹏錁涸ח铡僇דֹ׷ ➙תד鋅⳿ׁ׸גְזְ葺ְ㼎韋׾鋅⳿ׅ Ԩ ְ׆׸׮չ堣唒㷕统պ׌ֽדכ鍑ֽזְֿה׾ת׆椚鍑ׅ׷ֿהկ Ԩ ְ׆׸׮✮庠䪮遭ד֮׷堣唒㷕统ךأ؝٦ف㢩ך㉏겗ד֮׶ծ堣唒㷕统 ⟃㢩ך׮ך ➜Ⰵ㹋꿀ծ㼔Ꟍ涸濼鋅ծ㼔Ꟍ㹺הך⼿⫴ ָ䗳갭 Ԩ չ➂䊨濼腉 "* պה鎉ֲה傀ח鍑寸䪮遭ָ֮׷״ֲז孡ָ׃ג׃תֲָ 劢鍑寸㉏겗կ猰㷕罏ה䠐陎ך굸ְ麩ְָ饯ֿ׶װֻׅ湱✼椚鍑ָꅾ銲 Ԩ ➂䊨濼腉ⴓꅿדכչ涪鋅׾荈⹛⻉דֹ׷ךַպכꅾ銲ז劢鍑寸铬겗 ̔猘׋׍כ傈ղչ涪鋅պהչ㷕统պ׾粸׶鵤׃ג欰ֹגְ׷ ך麓玎׾➂꟦ךꣲ׵׸׋钠濼腉⸂ך眔㔲ד椚鍑ׅ׷ x!y AAACBXicbVC7SgNBFL0bXzG+opY2E4NgFXZF0DJoYxnBPCC7htnJJBkys7PMzIrLktpfsNXeTmz9Dlu/xEmyhSYeuHA4517O5YQxZ9q47pdTWFldW98obpa2tnd298r7By0tE0Vok0guVSfEmnIW0aZhhtNOrCgWIaftcHw99dsPVGkmozuTxjQQeBixASPYWOneD0X2OPErvpF+Je2Vq27NnQEtEy8nVcjR6JW//b4kiaCRIRxr3fXc2AQZVoYRTiclP9E0xmSMh7RraYQF1UE2+3qCTqzSRwOp7EQGzdTfFxkWWqcitJsCm5Fe9Kbiv14oFpLN4DLIWBQnhkZkHjxIODISTStBfaYoMTy1BBPF7O+IjLDCxNjiSrYUb7GCZdI6q3luzbs9r9av8nqKcATHcAoeXEAdbqABTSCg4Ble4NV5ct6cd+djvlpw8ptD+APn8wfECJj5 AAACBXicbVC7SgNBFL0bXzG+opY2E4NgFXZF0DJoYxnBPCC7htnJJBkys7PMzIrLktpfsNXeTmz9Dlu/xEmyhSYeuHA4517O5YQxZ9q47pdTWFldW98obpa2tnd298r7By0tE0Vok0guVSfEmnIW0aZhhtNOrCgWIaftcHw99dsPVGkmozuTxjQQeBixASPYWOneD0X2OPErvpF+Je2Vq27NnQEtEy8nVcjR6JW//b4kiaCRIRxr3fXc2AQZVoYRTiclP9E0xmSMh7RraYQF1UE2+3qCTqzSRwOp7EQGzdTfFxkWWqcitJsCm5Fe9Kbiv14oFpLN4DLIWBQnhkZkHjxIODISTStBfaYoMTy1BBPF7O+IjLDCxNjiSrYUb7GCZdI6q3luzbs9r9av8nqKcATHcAoeXEAdbqABTSCg4Ble4NV5ct6cd+djvlpw8ptD+APn8wfECJj5 AAACBXicbVC7SgNBFL0bXzG+opY2E4NgFXZF0DJoYxnBPCC7htnJJBkys7PMzIrLktpfsNXeTmz9Dlu/xEmyhSYeuHA4517O5YQxZ9q47pdTWFldW98obpa2tnd298r7By0tE0Vok0guVSfEmnIW0aZhhtNOrCgWIaftcHw99dsPVGkmozuTxjQQeBixASPYWOneD0X2OPErvpF+Je2Vq27NnQEtEy8nVcjR6JW//b4kiaCRIRxr3fXc2AQZVoYRTiclP9E0xmSMh7RraYQF1UE2+3qCTqzSRwOp7EQGzdTfFxkWWqcitJsCm5Fe9Kbiv14oFpLN4DLIWBQnhkZkHjxIODISTStBfaYoMTy1BBPF7O+IjLDCxNjiSrYUb7GCZdI6q3luzbs9r9av8nqKcATHcAoeXEAdbqABTSCg4Ble4NV5ct6cd+djvlpw8ptD+APn8wfECJj5 AAACBXicbVC7SgNBFL0bXzG+opY2E4NgFXZF0DJoYxnBPCC7htnJJBkys7PMzIrLktpfsNXeTmz9Dlu/xEmyhSYeuHA4517O5YQxZ9q47pdTWFldW98obpa2tnd298r7By0tE0Vok0guVSfEmnIW0aZhhtNOrCgWIaftcHw99dsPVGkmozuTxjQQeBixASPYWOneD0X2OPErvpF+Je2Vq27NnQEtEy8nVcjR6JW//b4kiaCRIRxr3fXc2AQZVoYRTiclP9E0xmSMh7RraYQF1UE2+3qCTqzSRwOp7EQGzdTfFxkWWqcitJsCm5Fe9Kbiv14oFpLN4DLIWBQnhkZkHjxIODISTStBfaYoMTy1BBPF7O+IjLDCxNjiSrYUb7GCZdI6q3luzbs9r9av8nqKcATHcAoeXEAdbqABTSCg4Ble4NV5ct6cd+djvlpw8ptD+APn8wfECJj5 ׾ⵃ欽׃ג葺ְ׾䭯א׾涪鋅ׅ׷ x AAAB/XicbVDLSgMxFL2pr1pfVZdugkVwVWZE0GXRjcsK9gHtUDJppo1NMkOSEctQ/AW3uncnbv0Wt36JaTsLbT1w4XDOvZzLCRPBjfW8L1RYWV1b3yhulra2d3b3yvsHTROnmrIGjUWs2yExTHDFGpZbwdqJZkSGgrXC0fXUbz0wbXis7uw4YYEkA8UjTol1UrMbyuxx0itXvKo3A14mfk4qkKPeK393+zFNJVOWCmJMx/cSG2REW04Fm5S6qWEJoSMyYB1HFZHMBNns2wk+cUofR7F2oyyeqb8vMiKNGcvQbUpih2bRm4r/eqFcSLbRZZBxlaSWKToPjlKBbYynVeA+14xaMXaEUM3d75gOiSbUusJKrhR/sYJl0jyr+l7Vvz2v1K7yeopwBMdwCj5cQA1uoA4NoHAPz/ACr+gJvaF39DFfLaD85hD+AH3+ADzJlfc= AAAB/XicbVDLSgMxFL2pr1pfVZdugkVwVWZE0GXRjcsK9gHtUDJppo1NMkOSEctQ/AW3uncnbv0Wt36JaTsLbT1w4XDOvZzLCRPBjfW8L1RYWV1b3yhulra2d3b3yvsHTROnmrIGjUWs2yExTHDFGpZbwdqJZkSGgrXC0fXUbz0wbXis7uw4YYEkA8UjTol1UrMbyuxx0itXvKo3A14mfk4qkKPeK393+zFNJVOWCmJMx/cSG2REW04Fm5S6qWEJoSMyYB1HFZHMBNns2wk+cUofR7F2oyyeqb8vMiKNGcvQbUpih2bRm4r/eqFcSLbRZZBxlaSWKToPjlKBbYynVeA+14xaMXaEUM3d75gOiSbUusJKrhR/sYJl0jyr+l7Vvz2v1K7yeopwBMdwCj5cQA1uoA4NoHAPz/ACr+gJvaF39DFfLaD85hD+AH3+ADzJlfc= AAAB/XicbVDLSgMxFL2pr1pfVZdugkVwVWZE0GXRjcsK9gHtUDJppo1NMkOSEctQ/AW3uncnbv0Wt36JaTsLbT1w4XDOvZzLCRPBjfW8L1RYWV1b3yhulra2d3b3yvsHTROnmrIGjUWs2yExTHDFGpZbwdqJZkSGgrXC0fXUbz0wbXis7uw4YYEkA8UjTol1UrMbyuxx0itXvKo3A14mfk4qkKPeK393+zFNJVOWCmJMx/cSG2REW04Fm5S6qWEJoSMyYB1HFZHMBNns2wk+cUofR7F2oyyeqb8vMiKNGcvQbUpih2bRm4r/eqFcSLbRZZBxlaSWKToPjlKBbYynVeA+14xaMXaEUM3d75gOiSbUusJKrhR/sYJl0jyr+l7Vvz2v1K7yeopwBMdwCj5cQA1uoA4NoHAPz/ACr+gJvaF39DFfLaD85hD+AH3+ADzJlfc= AAAB/XicbVDLSgMxFL2pr1pfVZdugkVwVWZE0GXRjcsK9gHtUDJppo1NMkOSEctQ/AW3uncnbv0Wt36JaTsLbT1w4XDOvZzLCRPBjfW8L1RYWV1b3yhulra2d3b3yvsHTROnmrIGjUWs2yExTHDFGpZbwdqJZkSGgrXC0fXUbz0wbXis7uw4YYEkA8UjTol1UrMbyuxx0itXvKo3A14mfk4qkKPeK393+zFNJVOWCmJMx/cSG2REW04Fm5S6qWEJoSMyYB1HFZHMBNns2wk+cUofR7F2oyyeqb8vMiKNGcvQbUpih2bRm4r/eqFcSLbRZZBxlaSWKToPjlKBbYynVeA+14xaMXaEUM3d75gOiSbUusJKrhR/sYJl0jyr+l7Vvz2v1K7yeopwBMdwCj5cQA1uoA4NoHAPz/ACr+gJvaF39DFfLaD85hD+AH3+ADzJlfc= x!y AAACBXicbVC7SgNBFL0bXzG+opY2E4NgFXZF0DJoYxnBPCC7htnJJBkys7PMzIrLktpfsNXeTmz9Dlu/xEmyhSYeuHA4517O5YQxZ9q47pdTWFldW98obpa2tnd298r7By0tE0Vok0guVSfEmnIW0aZhhtNOrCgWIaftcHw99dsPVGkmozuTxjQQeBixASPYWOneD0X2OPErvpF+Je2Vq27NnQEtEy8nVcjR6JW//b4kiaCRIRxr3fXc2AQZVoYRTiclP9E0xmSMh7RraYQF1UE2+3qCTqzSRwOp7EQGzdTfFxkWWqcitJsCm5Fe9Kbiv14oFpLN4DLIWBQnhkZkHjxIODISTStBfaYoMTy1BBPF7O+IjLDCxNjiSrYUb7GCZdI6q3luzbs9r9av8nqKcATHcAoeXEAdbqABTSCg4Ble4NV5ct6cd+djvlpw8ptD+APn8wfECJj5 AAACBXicbVC7SgNBFL0bXzG+opY2E4NgFXZF0DJoYxnBPCC7htnJJBkys7PMzIrLktpfsNXeTmz9Dlu/xEmyhSYeuHA4517O5YQxZ9q47pdTWFldW98obpa2tnd298r7By0tE0Vok0guVSfEmnIW0aZhhtNOrCgWIaftcHw99dsPVGkmozuTxjQQeBixASPYWOneD0X2OPErvpF+Je2Vq27NnQEtEy8nVcjR6JW//b4kiaCRIRxr3fXc2AQZVoYRTiclP9E0xmSMh7RraYQF1UE2+3qCTqzSRwOp7EQGzdTfFxkWWqcitJsCm5Fe9Kbiv14oFpLN4DLIWBQnhkZkHjxIODISTStBfaYoMTy1BBPF7O+IjLDCxNjiSrYUb7GCZdI6q3luzbs9r9av8nqKcATHcAoeXEAdbqABTSCg4Ble4NV5ct6cd+djvlpw8ptD+APn8wfECJj5 AAACBXicbVC7SgNBFL0bXzG+opY2E4NgFXZF0DJoYxnBPCC7htnJJBkys7PMzIrLktpfsNXeTmz9Dlu/xEmyhSYeuHA4517O5YQxZ9q47pdTWFldW98obpa2tnd298r7By0tE0Vok0guVSfEmnIW0aZhhtNOrCgWIaftcHw99dsPVGkmozuTxjQQeBixASPYWOneD0X2OPErvpF+Je2Vq27NnQEtEy8nVcjR6JW//b4kiaCRIRxr3fXc2AQZVoYRTiclP9E0xmSMh7RraYQF1UE2+3qCTqzSRwOp7EQGzdTfFxkWWqcitJsCm5Fe9Kbiv14oFpLN4DLIWBQnhkZkHjxIODISTStBfaYoMTy1BBPF7O+IjLDCxNjiSrYUb7GCZdI6q3luzbs9r9av8nqKcATHcAoeXEAdbqABTSCg4Ble4NV5ct6cd+djvlpw8ptD+APn8wfECJj5 AAACBXicbVC7SgNBFL0bXzG+opY2E4NgFXZF0DJoYxnBPCC7htnJJBkys7PMzIrLktpfsNXeTmz9Dlu/xEmyhSYeuHA4517O5YQxZ9q47pdTWFldW98obpa2tnd298r7By0tE0Vok0guVSfEmnIW0aZhhtNOrCgWIaftcHw99dsPVGkmozuTxjQQeBixASPYWOneD0X2OPErvpF+Je2Vq27NnQEtEy8nVcjR6JW//b4kiaCRIRxr3fXc2AQZVoYRTiclP9E0xmSMh7RraYQF1UE2+3qCTqzSRwOp7EQGzdTfFxkWWqcitJsCm5Fe9Kbiv14oFpLN4DLIWBQnhkZkHjxIODISTStBfaYoMTy1BBPF7O+IjLDCxNjiSrYUb7GCZdI6q3luzbs9r9av8nqKcATHcAoeXEAdbqABTSCg4Ble4NV5ct6cd+djvlpw8ptD+APn8wfECJj5

Slide 216

Slide 216 text

/ 166 猰㷕涸椚鍑הכ荈搫猰㷕灇瑔כ➂꟦ך㌀׫ד֮׷ 157 Ԩ 荈搫ך׃ֻ׫׾ չ➂꟦ָպ椚鍑ׅ׷䗳銲ָ֮׷ չ椚鍑պחꟼ⤘ׅ׷灇瑔װ䪮遭ָ♧瘡篖דכְַזְךכׅץגֿךְׇ ˖ 钠濼ꣲ歲➂꟦כ㣐ꆀך䞔㜠׾椚鍑דֹזְ٥䙼ְ鴥׫ ⟎铡 ָ䗳갭 ˖ 钠濼غ؎،أ٥爡⠓غ؎،أ➂꟦ָر٦ة׾ה׷הغ؎،أכ♶〳鼘 ˖ 䠬䞔 䗰椚 猘׋׍ך䙼罋٥ⴻ倖כ䠬䞔ח鏮ִ׷倯岀װ䞔㜠乼⡲ח腚ְ ˖ 猰㷕涸椚鍑ך䧭⸆غ؎،أ猰㷕涸椚鍑כ⳿晛ׁ׸גכׄ׭ג⚅ך⚥ח 䎢ת׶涺ך猰㷕涸椚鍑הז׷ָծ⳿晛穠卓כ溪㹋ך♧鿇ך׫ח䓼ֻ⨉׷ ˖ ת׋ծ荈搫ך岀⵱ָ䗳׆׃׮➂꟦ך׃׳רְ钠濼ꣲ歲ך眔㔲ד知患ח鎸 鶢׃׋׶ⵖ䖴דֹ׷הכꣲ׵זְ Ԩ 䩛חⰅ׷䞔㜠כ䌢ח鿇ⴓ涸➂欰ָ剣ꣲד֮׷⟃♳ծׅץג׾ٌرٕ⻉׃ ׋׶ծ֮׶ה֮׵ײ׷䞔㜠׾Ⰵ⸂㢌侧ח׃׋׶ׅ׷ֿהכ♶〳腉կֿך䠐 ㄂ד⡦׵ַךչ㷕统պװչ濼陎պכ♶〳鼘

Slide 217

Slide 217 text

/ 166 㹋꿀荈⹛⻉铩׌׏ג⽃锃ד鷌㾄ז⸤⫴ַ׵鍑佝ׁ׸׋ְ 158 Organic synthesis in a modular robotic system. Science 363 (2019) A mobile robotic chemist. Nature 583 (2020) Automating drug discovery. Nature Reviews Drug Discovery 17 (2018) Ԩ 猰㷕灇瑔חְֶג׮ꬊ⸬桦ז⸤⫴ְָ׆׸荈⹛⻉ׁ׸׷ךכ娖〷涸䗳搫 ̔猰㷕חאֹתהֲⱄ植䚍٥㾩➂䚍٥麓ꃎ⸤⫴ך㉏겗鍑嶊ח׮项ׅ׷

Slide 218

Slide 218 text

/ 166 ׋׌׃⡲噟ך荈⹛⻉ה涪鋅׉ך׮ךך荈⹛⻉כⴽ如⯋ך㉏겗 159 Ԩ ٗنحز禸ָ֮׸ל笺뜧ז㹋꿀ָ ➂䩛״׶כ ׋ֻׁ׿דֹ׷ָ׉׸ד׮ չ涪鋅պחכ⸂䪮ך⸇鸞⻉דכⵋ麦דֹזְ 〳腉ז⦪酡瑞꟦ָ䎢ֺׅ Ԩ 荈⹛⻉䪮遭ך灇瑔הさ׻ׇגչ⡦׾וֲ涪鋅ׅ׷ךַպך䨌殛ָ䗳갭 ̔䱱稊ׅ׷禸٥أ؝٦فծ堣㐻ד鎘庠ׅץֹ㢌侧ծةأؙךرؠ؎ٝ

Slide 219

Slide 219 text

/ 166 ➂䊨濼腉ⴓꅿחֶֽ׷չ涪鋅պך灇瑔 160 涪鋅猰㷕ծ猰㷕涸涪鋅ך锷椚ծ濼陎涪鋅ծ堣唒㷕统ծ،ـتؙءّٝ

Slide 220

Slide 220 text

/ 166 涪鋅猰㷕ה堣唒涪鋅 161

Slide 221

Slide 221 text

/ 166 涪鋅猰㷕ה堣唒涪鋅 162 խխ䎷畍խ傈劤ך涪鋅猰㷕فٗآؙؑز剣䊛眍㣗 痥*鿇խ䱿锷ח״׷濼陎涪鋅 խխ痥畍խչ涪鋅պך猰㷕ㆸ㷕ˌ娖〷涸稆䲽ꅿ㹺㉔♧ խխ痥畍խ窟鎘涸鎸〾Ⳣ椚鎉铂13*4.✡靼歋ꥐ٥⡟谏岱➜ խխ痥畍խ✮庠ٌرַٕ׵ךٕ٦ٕ䬄⳿ˌ侧䒭ַ׵鎉铂פ剢劤խ峔٥啾歊⼪窩 խխ痥畍խ䌓秛锷椚فؚٗٓىؚٝה鏾僇酡㸣㿊劤畍⽆٥剣募⽆秀٥䎂歊罭♧ խխ痥畍խ,FZ(SBQIˌؗ٦ٙ٦س䬄⳿خ٦ַٕ׵涪鋅خ٦ٕפך㾜Ꟛ瀧㿊խ床٥㣐愮䎊欰٥靼ⰻ歊姻䕕 խխ痥畍խ*OUFMMJHFOU1BEךさ䧭הⱄⵃ欽ˌ䌓秛䱿锷ך甧㜥ַ׵⾱〡խ铆٥䎂歊խ幄 痥**鿇խ鎘皾㷕统椚锷ח㛇בֻ濼陎涪鋅 խխ痥畍խ腉⹛㷕统ה涪鋅猰㷕㸜⦓湫埠٥꼛鋅㝩խ䬪 խխ痥畍խֻ׶鵤׃؜٦يה׃גך㷕统،ٕ؞ٔؤي⚵䀤խ畍٥戣劤薉✳ խխ痥畍խ؝ٝؾُ٦ة؟؎ؒٝأך׋׭ך⽃秪ַא⸬桦涸ז؟ٝفؚٔٝ䪮岀床鴟խ屚 խխ痥畍խ㷕统،ٕ؞ٔؤيך鐰⣣♳⾱齥僻 խխ痥畍խ䎗⡦ؙٓأةؚٔٝך䞔㜠鎘皾䎗⡦圓鸡➙❁խ嵞 խխ痥畍խ4VQQPSU7FDUPS.BDIJOFח״׷ⴓ겲넝갭幄㸩 痥***鿇խ堣唒㷕统הر٦ةو؎صؚٝח㛇בֻ濼陎涪鋅 խխ痥畍խ؝ٝؾُ٦ة佄䴂ח״׷猰㷕涸濼陎ך涪鋅1BU-BOHMFZ衼㹧ꅿխ䝎٥⚵㿊խ⥜鏬 խխ痥畍խ㷕统ַծو؎صַؚٝծٌرַؚٔٝˌ〢欰䡾㷕ַ׵ך✲⢽灇瑔)FJLLJ.BOOJMBFUBM衼啾♴溪♧鏬 խխ痥畍խⴓ協ꣲ㹀岀׾欽ְ׋⚛⴨ؚٓؿ䱱稊ח״׷剑黝穠さٕ٦ٕך涪鋅⚥靼僇䒽٥啾♴溪♧ խխ痥畍խ濼陎涪鋅ה荈䊹穈籼㘗ך窟鎘ٌرٕ⻌䊛彁㔊龤٥垌〡濼⛒ խխ痥畍խ곁㹏ך飑顠㾶娖ַ׵ךر٦ةو؎صؚٝ濶歊⹧⤥٥⸇谏湫埠٥纷㹓遤⥋ խխ痥畍խ涪鋅ءأذيהؼُ٦وٝؒؗأػ٦زך؎ٝذؚٖ٦ءّٝ⚵㿊խ⥜٥㹧ꅿխ䝎 痥*7鿇խ㣐鋉垷侧⦼ر٦ةַ׵ך濼陎涪鋅 խխ痥畍խ㣕ꤿ㖑椔禸暟椚㷕פך濼陎涪鋅ך䘔欽㹺啾⤥䕕٥♳ꅿ桠㣕٥腉⹲姻➋٥歕歊խ䗼٥虠加խ䗡٥✡❁鞢宕٥畾歊꧇䕕 խխ痥畍խـٓ؎ٝسإػٖ٦ءّٝהؐؑ٦ـٖحزח״׷ꦀ誊歗⫷ך涪鋅倜䃊罭♧ խխ痥畍խ鎘皾堣ח״׷猰㷕涸岀⵱٥ٌرٕך涪鋅倯岀ך㾜Ꟛ똲㽵խꥐ٥⯋歊խ嵞 խխ痥畍խ㢳㢌ꆀر٦ةַ׵ך㢳갪䒭㘗岀⵱ך涪鋅⚥ꅿ葺䎂٥俕谏ㄤ䊻 խխ痥畍խ갈㡮ر٦ةك٦أַ׵ך갈㡮濼陎ך涪鋅ꈿ加㛇⛒٥暜ꅿ姻♲ խխ痥畍խ⟎䟝⻉ׁ׸׋➂⡤ַ׵ךشؽ؜٦ءّٝח㛇בֻ濼陎涪鋅ך佄䴂خ٦ٕ뢷谏鞢俑٥둷腜秪♧龤 痥7鿇խطحزٙ٦ؙ橆㞮חֶֽ׷濼陎涪鋅 խխ痥畍խى٦يًر؍،׾欽ְ׋濼頿崧鸐ה猰㷕䪮遭ر٦ةך〳鋔⻉歊⚥խ隁 խխ痥畍խؤ٦ىؚٝ䪮遭׾欽ְ׋㼎鑧涸䞔㜠嗚稊؎ٝةؿؑ٦أ鞢歊姻〷٥叝㿊䝜ㆂ խխ痥畍խؙٔٝ䞔㜠ַ׵ך濼陎笨圓䧭䑜䊛⡟⼪歑٥寑歊㣐鰊٥歊〡ⶌ〷 խխ痥畍խ؎ٝة٦طحزדך⟰噟꟦䞔㜠Ⱏ剣חぢֽ׋وٕثؒ٦آؑٝزءأذي嬁ⵃꥐ㣗٥넝歊酔䘊 1BU-BOHMFZה)FJLLJ.BOOJMBך ꅾ銲锷俑ךㄤ鏬׮אְגְ׷

Slide 222

Slide 222 text

/ 166 邌植ה➜Ⰵ堣唒㷕统ַ׵堣唒涪鋅פ荚׷׋׭ך䭷ꆙ 163 Ԩ 黝ⴖזⰅ⸂邌植ךرؠ؎ٝה㷕统 ⴓ㶨邌植ה(//Tծ暴ח✲⵸㷕统ה׉ך鯄獳 Ⰵ⸂暴䗙ꆀ 鎸鶢㶨 ך鏣鎘٥ؒٝآص،ؚٔٝ Ԩ 堣唒㷕统ٌرٕ Ⰵ⳿⸂وحؾؚٝ ך邌植 椚锷鎘皾堣唒㷕统 ؚٖ؎نحؙأ剑黝⻉ ➜Ⰵ湱ꟼַ׵㔓卓ך呎䬿׾䖤׷ 邌植䗳銲⼧ⴓז䞔㜠׾堣唒חⰅ׸׷ Ԩ 湱ꟼ׾㔓卓ה然鏾ׅ׷׋׭ך➜Ⰵ㹋꿀灇瑔 䱱稊هٔء٦ה黝䘔涸㹋꿀鎘歗ךرؠ؎ٝ Ԩ ءىُٖ٦ءّٝװ㹋꿀荈⹛⻉הך輐さ

Slide 223

Slide 223 text

/ 166 堣唒㷕统٥堣唒涪鋅חה׏ג׮㹋⚅歲嗚鏾ך׋׭ך咿㿊屯 164 涪鋅٥椚鍑 邌植 ➜Ⰵ ⟎铡 An exciting “real-world” test bench for ML researchers! Ԩ 堣唒㷕统 .BDIJOF-FBSOJOH ꨄ侔圓鸡穈さׇ圓鸡׾⠵ֲ堣唒㷕统 ⴓ㶨ծ⿾䘔ծ⿾䘔穗騟/FUXPSL Ԩ 堣唒涪鋅 .BDIJOF%JTDPWFSZ 椚锷鎘皾.- ؚٖ؎نحؙأ剑黝⻉ ծ㹋꿀.- 䱱稊ה➜Ⰵ鎘歗 䖤׵׸גְ׷庠㹀ر٦ة 傀濼ך✲㹋 俑柃ر٦ة װ侄猰剅涸濼陎 ✲⵸䞔㜠 ٌرٕ 倜׋ז䞔㜠 ˖ 剣⸬זⰅ⸂㢌侧ךず㹀 ˖ إحز،حفךرؠ؎ٝ ˖ 㹋꿀ך鎘歗٥㹋倵 ˖ 穠卓ך鐰⣣٥鍑ꅸ ⟎铡 ٌرٕ ✲⵸䞔㜠

Slide 224

Slide 224 text

/ 166 ֿךأٓ؎س׮縧ְג֮׶תׅ 165 https://www.slideshare.net/itakigawa/presentations https://itakigawa.github.io/news.html

Slide 225

Slide 225 text

/ 166 תה׭➙傈ך鑧 166 ˖ 荈䊹稱➜ 堣唒㷕统ה荈搫猰㷕ך㞮歲 ˖ 堣唒㷕统הכ倜׃ְفؚٗٓىؚٝך倯岀 ˖ 堣唒㷕统㾊כ♧⡤⡦ָ嚂׃ְךַ ˖ ⴓ㶨ך邌植ה堣唒㷕统 ˖ ؚٖ؎نحؙأ剑黝⻉ 怴糊䌓秛 锷椚㷕ה窟鎘㷕ך輐さ ˖ 荈搫猰㷕灇瑔ד堣唒㷕统׾⢪ֲֶהׅ׷ה䗳׆עאַ׷劤䔲חꨇ׃ְ㉏겗 ˖ ر٦ةٌرؚٔٝה✮庠،ٕ؞ٔؤي 5IF5XP$VMUVSFT ˖ ✮庠ַ椚鍑ַ3BTIPNPO⸬卓 6OEFSTQFDJDBUJPO 鍑ꅸ㢳圫䚍 ˖ ➂꟦ך钠濼غ؎،أח歋勻ׅ׷㉏겗⟎铡ծ㣟侁ծ䧭⸆غ؎،أծFUD ˖ 堣唒㷕统ַ׵堣唒涪鋅פ ˖ չ涪鋅պչ椚鍑պך麣瘡כさ椚⻉דֹ׷ךַ荈⹛⻉דֹ׷ךַ https://itakigawa.github.io/data/talk_20211026.pdf