MD-DSC研究会講演資料:『機械学習コンペティションの実際とその意義』/ Practice on ML competition and its significance

2019年1月22日 医療・創薬 データサイエンスコンソーシアム研究会 講演資料

Transcript

1. -) M S C AI : a 01 /01 21 ( c 9 ( A C D 9 9
2. h a ( 21 1- 40 o( c A 76: 6:7 ) ) M 7 r ( z ( k I i u 7 s
3. 
       0 S 234 43 1 S e 1 S e 1 S e 7 T 1 S e An J 6 DS GECJ 5A J AI o BE C P DS
4. n 4 n n
5. ). 4 421 34 ( 42 42 5 42 !#
   " $  #    
6. "53 4("))
   6 
   -2*
   :1#'$&/,  (96
   .+ !%7 804  
7. 7 n G l e n 11 y G n s rt ce H S 7f R e o S n 7 l g 2 1 R7 g p S l n A f S 1 0 10R d 7 n Q i S G
8.    
   n A DK
9. n A B I 4 n 4 O ( ) ONP4 Virtual screening … +(&,' !
   
   
   !1 !2 !3 $  +(&,' " %!* # ) screening 
   * # virtual screening
10. n K
     967 ( #  &).0 *! "%*@ =1;=>*  *(400) 4?:<A5).0 “”* */$, 862?3'  +(-OK
11. I T n 0,.. BD C7B99 icf [ L ? 1 20 N R ( ? F ,. 7 )7 0: : 0 . 0 R ( ] YSe y [ af ] ] : 9B B D ultm h s ui gi frns ui I ] ] [ vp [d [ Mg I 1 DC ? 2 C 8 ( ) ) ( 5
12. ( 7 2 Extended Connectivity Fingerprint Functional Connectivity Fingerprint Topological Torsions Atom Pairs Fingerprint RDKit Fingerprint Avalon fingerprint  !fingerprint
     (6) 70 Random Forest Extremely Randomized Trees Gradient Boosted Trees Multilayer perceptron Support Vector Regression  ! $)%(&'
     (5) 65 = 30# Elastic Net Pfinal Level 0 Level 1 "  #  1 2 ) ):
13.  Fingerprint ECFP FCFP TT AP RDK AVLN F-Stacking RF 0.848 0.855 0.816 0.686 0.652 0.722 0.892 ERT 0.869 0.889 0.844 0.798 0.671 0.768 0.907 GBT 0.852 0.864 0.835 0.808 0.733 0.758 0.891 MLP 0.802 0.777 0.623 0.814 0.651 0.712 0.895 SVR 0.856 0.852 0.688 0.763 0.662 0.693 0.877 L-Stacking 0.890 0.911 0.870 0.881 0.799 0.846 0.930 FL-Stacking   Level0
     ROC-AUC  (3 )1)072 n 0 3 0 0
14. 1 n o y mls))(,D ipF e n )- JA H DN TB a e a n AB a e R )-0 1 18 2 6 1 - 6 1 0 1 r tqR ni +*,
    )('!  #% )('! 
    " $&    +*,
15. 1 5 n C ( 2 N3 8B? T DG 1 ) ) )1 1 ) ) ) ) 1 1 ) ) ) ) )1 1 1 1 1 1 1 6 A B 0 T DG 1 ) ) ) ) ) 1 1 ) ) ) 1 1 ) ) ) ) ) )1 1 1 1 1 1 1 0 J J :
16. )1 ( 6 1 3 n WR N TF.::A, c n TF 22 a .2 32 iJ M eb gd h , A .: 1 5: B D8 , 0 /B: 2 CDB: C:8 ) (
17. B (A N 7 /: 7 n 5,-* 3/
    %' 0)!(%'23/
    "1470)6 .#$+#  &  ) D 7 /: 7 7 /: 7J 1
18. 1 8 B F J D n 0. 1 1 F : D B A 1 0 0 0 0.00 0.00 C 0 1 0 0 1.00 0.00 T 0 0 0 1 0.99 0.00 T 0 0 0 1 0.99 0.00 … G 0 0 1 0 0.41 0.00 A 1 0 0 0 0.00 0.00 ()
19. 1 ( : 2 DJ B ) 9 n t i D c Ai D A A he N D ? ? rs Aam n t ul o u Ai ? n D ( ( ) A Aam
20. : B 2 n A- U O10 26 . 0 n 2 . 8 0 ( J D : ) : (0)2 : 4 C A R
21. 9B 00.1 5 AB: D 6 O C D J C D C D ia k S 98 32 -4 ) ( n C - D n cg 3 ia k n B e N d n
22.      
      n - 2 i I a v L ( we IT s I . 0 228 8 c n ) ) uhn yp n c bD I n log P r c bD t I d k( : ( . 0 .8 / 11 . 1 2
23. 21 n ( e n a K l n 3 ) g )
24. 4 2 n 3 t 1 o r ru 2 a i 1 o ru ) 2 l1 ru 1es 2 1 r f K2 n (( h g g K2
25. 5 5 n 2 5 ( ) ) n : )
26. 1 2 6 1 3 .(2 n 0 n 0 3 ) 2 n 0 0 1 1 3 n n 0 1 2
27.