AIアプリはこう作る！-独自の識別モデル作成も簡単 Einstein Platform Services の使い方
4 likes2,144 views
Einstein Platform Services provides capabilities for vision, language, and lightning components. It offers image classification, object detection, intent analysis, and Apex development. Documentation is available on setting up and calling APIs through Node.js, Heroku, and lightning components.
![kg up [c W Ud 8 ( . kT m
T ag msT bR ToTa fi
CG L CG E L HJF JNC
HH
.F A
)E CP J
G J E
.F A
)E CP J
npsa
r
a
r
( .ah t g s
LLI IC CG L CG C vTl : NC CHG t:
ihms
S2( L sTi
S rTo
S r
|m m
S / 21
JH b 0CA LGCGA E L HJF
.] df e
fi CG L CG E L HJF JNC
O E HJ](https://image.slidesharecdn.com/devwebinareinsteinplatformservices040518-180409015512/85/AI-Einstein-Platform-Services-8-320.jpg)
AIアプリはこう作る！-独自の識別モデル作成も簡単 Einstein Platform Services の使い方
- 1. 4 2 0 Einstein Platform Services (Vision/Language) 1 58 | Platform Specialist
- 2. (2018 4 5 ) o| 0 S 8 2S a o | 2 8 a: https://einstein.ai/ r f fce 1 1 2 8 fce s " " 2S l 0 c 8 S
- 3. r t N J 2 E t • | R g jod U 1. Pa • 2 E . FA A 0 F U a • H E F / 8 E H 0 0 1 I a N • ) AEF A 0 F 2 H E (0 sSyc STL u mhpc e od jfnl i 3 8 cv ba a
- 4. S(T C C ) Einstein Platform Services e• K• C O Vision• - Image Classification Vision• - Object Detection Language• - Intent Lightning Component & APEX• Node.js (on Heroku)• 1 2 3 üb a ü c 8 200 1 4 2
- 5. . .1
- 6. • Ø 8 28 0 • Ø 16 8 • Ø 8 • Ø 8 2
- 7. 3 b I A 8 , I7P ea fA ea W c S S 0 3 12
- 8. kg up [c W Ud 8 ( . kT m T ag msT bR ToTa fi CG L CG E L HJF JNC HH .F A )E CP J G J E .F A )E CP J npsa r a r ( .ah t g s LLI IC CG L CG C vTl : NC CHG t: ihms S2( L sTi S rTo S r |m m S / 21 JH b 0CA LGCGA E L HJF .] df e fi CG L CG E L HJF JNC O E HJ
- 9. s aZ I i |Z PS )( ( ( ( " Z e c i |" Z 48" PS I 4 24 .4 4" I f aZ i |" Z p ro" t E PS p 48" PS I 4 24 .4 4" I f aZ i |Z V aZ nl aZ i |Z p ro t E PS ) ( 0 2 3 912
- 10. SE fhES I I C e I S 0 8 0 2 . 1 0 80 g 0 8 0 S SE a i cE I i /88 0 0 8 0 0 0. : 1
- 11. Heroku Einstein Platform Services c gl i |vR P Hc gl( s ) 8 C lo u A 8 2 A i |v s 8 C lob dn uR lo Y u S lob dn K08A C 8 fk H " A eadn A 8 2 A Ir R K1 8 " 88 r K D A f aoga R K18 A fk ( F 2 8A eadn K08D8 ( F 2 A f aogaR tV -) E 1 8A 8
- 14. Ll Lte( Og L ) . 42 8 3 1 Lsi rsiRjn O 0 02 8 3 2 . 0 21 2 21 ouL D Rka O 0 02 2 21 4 42 2 cf Im LS V L Cb LDRjn O 2 4 8 3 2 - 82 3 12
- 15. t cs x Z / y S / X / { mdla ( ( ) ) 458 , 4 545 ," 5 7 : , 2 , v I Z rbd Z ) rbd . B 8 0 4 9 18 E 68B 3 0 S o f ni P gufp ed " 14 8B9 68
- 16. Vision - Image Classification f tp i SRg p or pLr ) 2 C 3p N Pk o nvp p p 30-p 8p 26 26 3p U kp lpZ ) o nv xm qs ) 1 a N cj ) i kT f tp p ( f p e E 2 3
- 17. l 8 1 e s c x n ls q g LLH ) MJD BI 0 g rv n T g n T ed c ls q g LLH ) L D AL M: A BI A A C xg e n ls q g LLH ) OOO A A C J JAHL A H H 2 g h . :J O kgq v U g oq f N JL g oq f l JA s f - s xwy g py kT jcgs n a g x n mmi b ml .721 | u / : gt LLH ) A : AL M: A A L A NA A R S ( D J
- 18. STEP1: u sf ) zx cd wo tl a u swmn l * u swmn l 2 53 ' qnr' f F % M I % S 0.( 0.( %J E M /- $ 684 u swmn l | 0( |ie| F % M M 0( $ 684 bX g f G G %G F $ 684 1 G M %M I 4 9 E P 6841 G - ; Iwmn f • f fpy v 2• l ` f u sf Ze e a| )/ *()/ 9 E
- 19. STEP2: Zipk am eln i(g dceh s) > 94: ru HTTPS BPI"EINSTEIN"BI TOKEN & 0<8 SBGE qof h CU @ 4> 0UTHO IXBTION 1EB E > 94:. HTTPS BPI"EINSTEIN"BI & BPIUSBGE AIPk am eln i CU @ < => 0UTHO IXBTION 1EB E > 94:. 2ONTENT > PE U TIPB T FO DBTB DBTB-/ AIPk am j k amt. T PE-I BGE HTTPS BPI"EINSTEIN"BI & ISION DBTBSETS UP OBD g dcehpsyz v CU @ 4> 0UTHO IXBTION 1EB E > 94:. HTTPS BPI"EINSTEIN"BI & ISION DBTBSETS g dceh83. Z& =B ESFO CE
- 20. STEP3: | y p( v ) | y p DRO B =< @ 0R H OIUC I L 1FCOFO @< . 2 L FL @TNF KR INCO G OK EC C LCKF- v f . EC CPF 8E- v s rt 83. H NP CNI FILP FIL CI S( SIPI L OCIL ( DRO B @ 0R H OIUC I L 1FCOFO @< . H NP CNI FILP FIL CI S( SIPI L OCIL v 83. ) u " DRO B =< @ 0R H OIUC I L 1FCOFO @< . 2 L FL @TNF KR INCO G OK EC C PCKN F2 L FL -/ l hfz l . K EF 8E- v 83. H NP CNI FILP FIL CI S( SIPI L NOFEID • )cX v s X n f l a c d A>: c i v srt• f f eX v 83cf c i m l o f 0=8 | H NP KF CKILE OFCEKF I E DP NOFEID ISF SIPI L PFOSIDF CNI" (& (& C FPG ODF
- 21. Vision - Image Classification s % e • % & & la s % • % e f & os ( | e • % os % | e • c M% s SM • x & B • % os &) M • % os % • os M • M r M B M B • B &% 8 & % /31 24501
- 22. Vision - Object Detection i k vNlZd No k c X j f p P : C 8: - 8 : 8Cp S n t sp v v (( xp eRTk v k u, 8 C - Cp 0 - t s N|r x eRTkN v 0 - 3: eRTkS m a 0 - l bnY i k v .E : C E 8) 1 2 .E : C E 8) 1 2 i k vT v ( 8C 8
- 23. b gec h P fa lf gec h P 2 A2A fa I 8AA 7 A8B3 B ( A 2 A2A 2A / A 2 /. U -1. i G I S 8AA 7 A A 2 ) 02:
- 24. STEP1: iCpt & r s h a m g i 8042 0 3 10 : pt e U aU l 8 0 5 ) :10 5: ) a r s Uh a / aUZ c a n• | f aU oh h G S o a Uh IS ( 0 2 3: 12
- 25. STEP2: Ziprce ks o(m ikn ) 2/ 1 GG 8 : G: 8 G : 3 GA8 f p 2/ 1 3 GA8 . II R | P 3. 8 : gt o { lhn 3 GA8 . m ikn 3 GA8 . 3 GA8• 3. 40 N R N L |} aTO S U { P ( K ) 58 : E :
- 26. STEP3: P ( I) I S 68 ( )C 68 ( )C P 68 ( )C ) 1 3 882 26• G P I U 3 806
- 27. Vision - Object Detection lfo ae f c • o • f c o| • o | M o B • o & S • x s M r o M • o • lfo 8 B M B 6 7 ( 1 023
- 28. Intent - Case Routing l o p f twE H Va n sP C | (3 13 2 1 3 1p SV rc p a c Va n S v E r o -p To o p ( mio ( - To S d ( cp er l o 0 )1 8 l o T I forgot my login password 1 2 1
- 29. ) 0 P ea 9 0: B 1 1 / AB )/. SVU 2. b G I T c :BB A( CCC 9 B AB E 0 A8
- 30. STEP1: 8 S 0 8) .1 8 .1 1 -: 8 :8 3 1 . ( 2.8/ .
- 31. STEP2: f c g 1. 0 E) AE A A 2 E A o in| 1. 0 N T 2 E kphT r eag T f c g T 2 E A• T 2 3/ K ml c OSKN I t LU | P RNT s rT G ( 4 E: 8
- 32. STEP3: S ( U ) U ) f cda ) f e I S ) f ? 8 31 1 3 10 8 31 (• b S U G C " 1 2 1
- 33. : S 8(Vision) 23 2 0 n r t s aS is S 23 2 0 t f s Ssc V | e S E oS 01 l ) 2 01
- 34. API k - 1-1 3 c |y sp Im • 3 c / E if |y sp Im P • |y sp Im a / E S R P e 0. / E o IT d |s • B 4 : "A 4 ": B A : : :B: B A : 4 : & n I tIr s • tIr Sh K • Iv n l p T R A P A A h K t f K P h tIr c K & 24 B8 A
- 35. 0 S 8 1 ( 2 35) 1 0
- 37. lsW W x( W ls) / A A C A A C I • 3 A . : 7 A / A G .A A / A A C A A S -2. L f biV| • 1 : 7 pj t lsV v aor C G "7 A :87 C 7 C I C CC 0 8 " • 3 A . : 7 A S eultU 2. W lscNe u • gnm r dS bhg SLaor C G "7 A :87 A A A C CC & & 1 : 7
- 38. Lightning Component o T bh Image Classification / Sentiment GUI W P ic fS 14 8C 8p y -8 8 A8 . D 2 4 84 0 4 : E r p Lgp DDAC : D E 4 4C 8AC 4A8 C4 A 84AA 38p in a vtsp reLp 12.0 y Txu 12.0 A8 ndT 12.0 / : D : A 8 D T | lLc 12.0( Lc o Lbmr T | ) ) 14 8C 8
- 39. Lightning Component S 1 f ( 1 1 l 2) 1 0 Lightning Component ( 98 3 e ) Apex ( 1 1 3 e ) 1 1 0 f 0 a 1 f e c Einstein Platform Services 1 o e
- 40. Lightning Component PAM e 8 1 1481 &81 1 4 2 1 1441 AS A cxd cxd cxd 1 0 S a l o| AS A A prxd l o|xd pr l o| ntxd ASsm ntxdnt nto| ASv C B" A H E f i " 41 2 1 1: https://developer.salesforce.com/docs/atlas.ja-jp.210.0.lightning.meta/lightning/controllers_server_actions_call.htm
- 41. STEP1: & or 0:8 : 8 S vs ne n .: A : 1 :4 : 2 C:4 A 1/ ih S A 3 : : A : 3: ( 0:8 : 8 mp cs o 0:8 : 8 | I ) g n g n 0:8 : 8• at f L I I | S mp cs o Se p d o v o S P L I mp cs o S S hl h v v at f v " )& E & 23 A 4
- 42. " "g ldis • {w} g ldis• 4D :F A5 EG F Pt• {w} Pt• rPn• | • 45 L • ruf}s xPl bacs Oy u b} T odpi rPn s {w} r} 1 rPn s {w} R {w} - DEF08 rPn s {w} S T 1 0 - 32 2 1 2 4D :F A C rPn se a {w} s 3D A8 IF rPn se a {w} s 4D :F A5 EG F58 IF M N " 8 E D:
- 43. STEP2: Apexe f u https://github.com/hinabasfdc/eps-lcapex-sampleapp01/tree/master/apex 13 8 08AC A" A A 8 4 18 8A |a 1. c ) 8E PN Pb e f 13) 8E 8 " A A 8 4 18 8A mvci J S t s c W _ 13) 8E 8 " A• nl 28 4 r e f xdR_ p o e f Oc ( 14 8A 8
- 44. STEP3: Lightning Component tn https://github.com/hinabasfdc/eps-lcapex-sampleapp01/tree/master/lc 4 tn m xp .0/ 2 4 tn 4 PfI UjpD 4A d _ Pf .0/ 4 8 4 4 tn r E D i l .0/ 4 4 8 4 tn eas c o g HUS U .0/ 34 C .1 4 24
- 45. STEP4: & S lf 534 8583 lf o lfCE 534 8583 818 o pe h ac nr 58 158 2 1 501 L h g & S lf lfCE 534 8583 is m P " o ( ) 1 2 01
- 46. STEP5: r bd cfm sI aPhw - / 1 6 0 C 3 & . 1 1 6 31 8 4 C " 1 6 3 "3 6 2 1 31 ": 0 8 1 1 1 1 4 . 1 4 1 j n a Pe pl n ot S gid v E & 01 6 3
- 47. Lightning Platform xPXe c I b CC 8:C D3 4 D A2 8 A 2 7 A4 : C : 2C7 A 2 • Pto a cPtoP /- z migS • 02 7 A4 1 P d O h CC 8:C D3 4 D A2 8 A 2 7 A4 : C : 3 4C C 4C: 3: 2 • .3 4C C 4C: 02 7 A4 a c b Ppl uO CC 8:C D3 4 2 4 : C : 2: • 2CC A O n r Is jf 02 7 A4
- 48. Node.js cgonPi a Image Classification T GUI p S kch Web d -8Cp Wk kchp EE D) : E 4 4D 8 DE 8 D D4 84 8p bfW .p ml 2 D 4 0E 8 01 / c V nVe n p ( H I ( 04 8D C 8
- 49. Node.js f e 1 aS ( 1 1 4 2) 1 S0 1 1 1 81 1 0 aS0 1 aS (9 4 2) c Einstein Platform Services
- 50. Node.js t socket.io E / r a E m ES m ES & 1 5 1 5 2 0 n o n c 1 P s m r fio r vo r fi v dl o r dl e n n 8 " 2
- 51. STEP1: c (Heroku d lS Gfe) Heroku d lH Github README d OK c d 5 1 3 5 0 1 : /0/ 421 3 3 /: 3/ & & -3 8 b h d oi 3 8 1 3/ 3 3 8 /22 1 3/ 3 3 3 / 3 ' c 5 /22 5 1 :: : 4 3 3/ 3 5 3 8 :/ 3 b gG m 3 8 3 d• oi -3 8 a nk & ./ 3 4 13
- 52. PS R N V / 8 E 1G5J .4 4 E 8 68 8 8 G 6 4 6 8E 8 E 8 E E4 8 6 8 .4 4 E : G5 6 G8 48:8 E4 8E 68 8 E 8 E 4 4 2 E : G5 6 G8 48:8 E4 8E 68 8 E 8 E E 0J E : G5 6 8 8 J4 E 8 3 E 0J 264 4 E : G5 6 4 88 8 E 8 E64 4 5 - E : G5 6 4 88 8 E 8 : 4 E # 24 8E 68
- 53. DEMO
- 55. S , 2 , 01 1 , 2 ! S 8 E A Pü ü 5 2 01
- 56. Trailhead Image Classification Intent d S / _ c aRe R A 5 1 5 1AA 1 8 A 1 851 A1 5A6 5 :1 :5 A 5 50 A 01 5E n hjPg 1 .5A C5 b e il 8 A 1 851 A1 5A6 5 :1 :5 A 2C 1 1 5A C5 1 81 5 5A 1 2 55 A A 5 5 - fm 8 A 1 851 A1 5A6 5 :1 C 5A 5 A 5 0 5 021A A I /1 5A6 5
- 57. Salesforce AI I 2018 4I 6 A, 5 0123 01 e I I 6 ) ) n ) ) E I ) 6 + 8 Dreamforce 2018 i ( ) ) http://bit.ly/appcontest_2018
- 59. C . . // : 4 4