Machine Learning の新機能✨ / What's New in Machine Learning

WWDC19で発表されたMLの新機能まとめスライドです。
このスライドを見ればWWDCのML関連の話題がだいたい掴めます。

- お断り -
このスライドは一般公開されているWWDC19のスライドのみ引用しています。

iOS de KANPAI !【WWDC 2019 報告会】
15 min

Transcript

1. .BDIJOF
   -FBSOJOH
   ͷ৽ػೳ✨
2. !LPPPPPUBLF ͡ΜΉ ͓அΓ
   ͜ͷεϥΠυ͸Ұൠެ։͞Ε͍ͯΔ88%$ͷεϥΠυͷΈҾ༻͍ͯ͠·͢
4. 
   ηογϣϯ
5. 7JEFP
   63-
   IUUQTEFWFMPQFSBQQMFDPNWJEFPTQMBZXXED 7JEFP ৄ͘͠஌Γ͍ͨ࣌͸
6. Summary Create ML Domain API Core ML 3 7JEFP
7. Create ML $SFBUF.- .-ϞσϧΛ4XJGUͰ࡞Δ͜ͱ͕Ͱ͖Δ
8. Create ML ࠓ
9. Sound Activity Tabular Image Text Create ML ৽
10. <9DPEF>
    
    <0QFO
    %FWFMPQFS
    5PPM>
    
    <$SFBUF
    .-> Create ML ৽
11. Create ML ৽ υϥοά
    
    υϩοϓͰ؆୯ʹ
12. Image Classifier Object Detector
13. %&.0
14. ෯޿͍.-ϞσϧΛ
    ؆୯ʹ࡞Δ͜ͱ͕Ͱ͖Δྫ Create ML
15. Image Classifier Object Detector 7JEFP
16. αΠίϩͷ໨ΛσΟςΫτ Dice Detection Model 5 5 4 4 6
17. Sound Classifier 7JEFP
18. ԻΛฉ͖෼͚Δ
19. Text Classifier Word Tagger 7JEFP
20. จষͷײ৘εύϜτϐοΫΛ෼ྨ Text Classification Topic Classification Spam/Not Spam Sentiment Analysis ! " Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Label 1 Label 2 Label 3 Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum
21. Tabular Classifier Tabular Regressor Recommender 7JEFP
22. Hardtrail Mountain Aspen Valley Pleasant Cove For You: Rocky Peak Cactus Ridge ͜ͷࢁΛߴධՁʹͨ͠ਓʹ
    ࣍ʹొΔࢁΛϨίϝϯυ
23. Activity Classifier 7JEFP
24. "QQMF
    8BUDI
    ͷηϯαΛར༻ Device Sensors g a r Accelerometer Gyroscope Magnetometer Altimeter
25. ͲͷߦಈΛ͍ͯ͠Δ͔Λ෼ྨ Jogging Standing Gestures Gaming Golf Swimming
26. .-
    Y
    "3 + 7JEFP
27. "QQMF
    1FODJM
    Ͱखॻ͖จࣈࣝผ
28. "3͢͝Ζ͘ͱ૊Έ߹Θͤ
29. ✨؆୯ʹ࣮ݱͰ͖Δ✨
30. ͡ΜΉͷਪ͠
    ϑϦεϏʔͷ౤͛ํ෼ྨͷ%&.0 Frisbee Motion Classifier Hammer Bowler Chicken Wing Backhand Forehand 7JEFP
31. Summary Create ML Domain API Core ML 3
32. %PNBJO
    "1* ֶशࡁΈͷϞσϧ
    
    ѻ͑Δ
33. Face capture quality Sound analysis Speech Saliency Speech on Mac Sentiment classification Animal Detection Text Recognition Object Tracking Attention Saliency Document Camera On device speech Word Tagging Image similarity Face Landmark Image Similarity Landmark Detection Image saliency Rectangle Detection Text catalog Word Embeddings Image Classification NL Transfer learning
34. 7JTJPO
    'SBNFXPSL
    *NBHF 7JEFP
35. ਓ͕Ͳ͜ʹண໨͢Δ͔Λֶश Attention and Objectness Based Saliency
36. લܠͱޙܠΛֶश Attention and Objectness Based Saliency
37. ˠը૾͔ΒॏཁͳΦϒδΣΫτΛநग़ Bounding Boxes (0,0)
38. ը૾ͷྨࣅ౓Λܭࢉ Precision and Recall Recall Percentage of Target Class retrieved from entire library
39. 'BDF
    -BOENBSLTͷվળ 76pt Face Landmarks 65pt Single Confidence Score Confidence Score per point New
40. ਓؒݕ஌ New Detectors - Human Detector
41. 
    
    New Detectors - Cat and Dog Detectors Cat Dog Dog Dog Cat
42. τϥοΩϯάਫ਼౓ͷվળ New Object Tracker
43. 7JTJPO
    'SBNFXPSL
    5FYU 7JEFP
44. จࣈ͓͜͠
45. ଎͞
    WT
    ਖ਼֬͞ Fast Versus Accurate Fast Accurate Processing time Optimized for real-time Asynchronous processing Memory footprint Smallest Larger Support for rotated text Limited Broad Support for variety of fonts Limited Diverse font styles Accuracy for natural language Good Best
46. ద੾ͳύϥϝʔλઃఆΛ͢΂͠ Use case • Read codes/serial numbers just like a barcode reader • Constrained camera usage • Interactivity is key request = VNRecognizeTextRequest(completionHandler: recognizeTextHandler) request.recognitionLevel = .fast
47. ӳޠͷΈ
    
48. /BUVSBM
    -BOHVBHF 7JEFP
49. จষͷײ৘ղੳ Natural Language Sentiment Analysis I was so excited for the season finale, $ Natural Language Sentiment Analysis but it was a bit disappointing. I was so excited for the season finale, $ ☹
50. Sentiment Analysis Text Classification Sentiment Analysis -1.0 1.0 Natural Language 0 “We had a not so fun time in Hawaii cause mom twisted her ankle.” -0.8 Natural Language Text ෛPSਖ਼ͷײ৘ͷ਺஋Խ
51. 4QFFDI
    BOE
    4PVOE 7JEFP
52. ΦϯσόΠεͰԻ੠ೝࣝ On-Device Device Support iPhone 6s and later iPad (5th generation) and later All
53. ೔ຊޠͳ͍ On-Device Language Support English
 United States, Canada, 
 Great Britain, India Spanish
 United States, Mexico, 
 Spain Italian Brazilian Portuguese Russian Turkish Chinese
 Mandarin and Cantonese
54. Ի੠ೝࣝ ˙ൃ࿩଎౓
    ˙Ұ࣌ఀࢭ
    ˙੠ͷಛ௃ ৽
55. ੠ͷಛ௃ ˙+JUUFS
    
    ໎͍ͷ͋Δ੠ʁ
    ˙4IJNNFS
    
    Ԓͷ͋Δ੠ʁ // Printing new results when recognizing pre-recorded audio if result.isFinal { let formattedString = result.bestTranscription.formattedString let speakingRate = result.bestTranscription.speakingRate let averagePauseDuration = result.bestTranscription.averagePauseDuration for segment in recognitionResult.bestTranscription.segments { let jitter = segment.voiceAnalytics?.jitter.acousticFeatureValuePerFrame let shimmer = segment.voiceAnalytics?.shimmer.acousticFeatureValuePerFrame let pitch = segment.voiceAnalytics?.pitch.acousticFeatureValuePerFrame let voicing = segment.voiceAnalytics?.voicing.acousticFeatureValuePerFrame } } ৽
56. Summary Create ML Domain API Core ML 3
57. $PSF.-
     7JEFP
58. Core ML 3 Model Flexibility Model
 Personalization NEW
59. Neural Network layers 100+ αϙʔτ
60. ഑෍ϞσϧϚγϚγ
61. ΦϯσόΠεΞοϓσʔτ Update Training examples UPDATED NEW
62. User Differences Ñ ൚༻తͳϞσϧͰ͸ਏ͍࣌
63. Update Task Ö ❤ á StickerClassifier Update Task CustomStickerClassifer UPDATED طଘͷϞσϧΛΞοϓσʔτ
64. On-Device Privacy Available No server ΦϯσόΠεͷྑ͞
65. 5VSJ
    $SFBUF 1ZUIPOͰ
    $PSF.-ϞσϧΛ࡞Δ͜ͱ͕Ͱ͖Δ
    ϥΠϒϥϦ 7JEFP
66. ୔ࢁͷը૾͕ඞཁ Original Approach ࠓ
67. Ұຕͷը૾Ͱ෺ମݕ஌ Synthetic Data Augmentation ৽
68. Synthetic Data Augmentation ৽ Ұຕͷը૾Ͱ෺ମݕ஌
70. 88%$ʹࢀՃ͢Δ͔೰ΜͰ͍Δਓ޲͚ IUUQTOPUFNVLPPPPPUBLFOOFBF
71. !LPPPPPUBLF ͡ΜΉ
72. %F/"͸ɺຊΠϕϯτͷ಺༰ɺฒͼʹ͓٬༷͕ຊΠϕϯτΛ௨ͯ͡ೖखͨ͠৘ใ౳ʹ͍ͭͯɺ
    ͦͷ׬શੑɺਖ਼֬ੑɺ࣮֬ੑɺ༗༻ੑ౳ʹ͖ͭɺ͍͔ͳΔ੹೚΋ෛΘͳ͍΋ͷͱ͠·͢ɻ