20190417 畳み込みニューラル ネットワークの基礎と応用

for 実践者向けディープラーニング勉強会 第二回
2019-04-17
https://dl4-practitioners.connpass.com/event/124498/

Transcript

1. 畳み込みニューラル
   ネットワークの基礎と応⽤
   実践者向けディープラーニング勉強会 第⼆回
   https://dl4-practitioners.connpass.com/event/124498/
   2019-04-17 (Wed) @ SB C&S
   

   View Slide

2. Agenda
   19:00 〜 19:05 opening
   19:05 〜 19:20 機械学習の導⼊
   19:20 〜 19:30 walkthrough Kerasを⽤いたコーディングサンプル
   19:30 〜 19:45 畳み込みニューラルネットワークの基礎
   19:45 〜 20:00 walkthrough CNNのフィルタの理解
   20:00 〜 20:10 休憩
   20:10 〜 20:25 breakout CNNアーキテクチャの紹介
   20:25 〜 20:40 Skymind Intelligent Layer (SKIL) の解説
   20:40 〜 20:55 walkthrough SKILを⽤いたモデルのデプロイ
   20:55 〜 21:00 closing
   !2
   

   View Slide

3. !3
   15
   

   View Slide

4. ニューラルネットワークの基礎
   !4
   

   View Slide

5. !5
   https://www.coursera.org/learn/introduction-tensorflow
   

   View Slide

6. Activity Recognition
   !6
   https://www.coursera.org/learn/introduction-tensorflow
   

   View Slide

7. Activity Recognition
   !7
   https://www.coursera.org/learn/introduction-tensorflow
   

   View Slide

8. Activity Recognition
   !8
   https://www.coursera.org/learn/introduction-tensorflow
   

   View Slide

9. Activity Recognition
   !9
   https://www.coursera.org/learn/introduction-tensorflow
   

   View Slide

10. !10
    https://www.coursera.org/learn/introduction-tensorflow
    

    View Slide

11. Activity Recognition
    !11
    https://www.coursera.org/learn/introduction-tensorflow
    

    View Slide

12. 機械学習の"Hello World”
    !12
    x = -2, -1, 0, 1, 2, 3, 4
    y = -3, -1, 1, 3, 5, 7, 9
    y = f(x)
    

    View Slide

13. 機械学習の"Hello World”
    !13
    x = -2, -1, 0, 1, 2, 3, 4
    y = -3, -1, 1, 3, 5, 7, 9
    y = f(x) = 2x + 1
    

    View Slide

14. 機械学習のアプローチ
    •適当にモデルを初期化 (y_=ax+b)
    !14
    

    View Slide

15. 機械学習のアプローチ
    • 適当にモデルを初期化 (y_=ax+b)
    •誤差（損失）を計算 (L=1/N Σ(y-y_)2)
    !15
    

    View Slide

16. 機械学習のアプローチ
    • 適当にモデルを初期化 (y_=ax+b)
    • 誤差（損失）を計算 (L=1/N Σ(y-y_)2)
    •誤差が⼩さくなるようにパラメータ (a, b)
    を少し更新 (a ← a - η ∂L/∂a)
    !16
    

    View Slide

17. 機械学習のアプローチ
    !17
    • 適当にモデルを初期化 (y_=ax+b)
    • 誤差（損失）を計算 (L=1/N Σ(y-y_)2)
    • 誤差が⼩さくなるようにパラメータ (a, b)
    を少し更新 (a ← a - η ∂L/∂a)
    •誤差（損失）を計算 (L=1/N Σ(y-y_)2)
    

    View Slide

18. 機械学習のアプローチ
    !18
    • 適当にモデルを初期化 (y_=ax+b)
    • 誤差（損失）を計算 (L=1/N Σ(y-y_)2)
    • 誤差が⼩さくなるようにパラメータ (a, b)
    を少し更新 (a ← a - η ∂L/∂a)
    • 誤差（損失）を計算 (L=1/N Σ(y-y_)2)
    •誤差が⼩さくなるようにパラメータ (a, b)
    を少し更新 (a ← a - η ∂L/∂a)
    • …
    

    View Slide

19. !19
    

    View Slide

20. !20
    Y
    ºB
    C
    Z@
    

    View Slide

21. !21
    

    View Slide

22. !22
    

    View Slide

23. 勾配降下法
    !23
    「⽬隠しで⾜元の勾配情報のみを使って⼭の頂上を⽬指すようなもの」
    学習率 η は歩幅のイメージ（η⼩＝すり⾜、η⼤＝巨⼈の⼀歩）
    IUUQTUXJUUFSDPNNPNJKJ@GVMMNPPOTUBUVT
    IUUQTXXXZBNBLFJPOMJOFDPNKPVSOBMEFUBJMQIQ JE
    

    View Slide

24. Fashion MNIST Dataset
    • 7万画像
    • 10カテゴリ
    • 28×28 pixels
    • 実験⽤データセット
    !24
    IUUQTHJUIVCDPN[BMBOEPSFTFBSDIGBTIJPONOJTU
    

    View Slide

25. !25
    

    View Slide

26. !26
    

    View Slide

27. !27
    

    View Slide

28. !28
    ʜ
    ʜʜʜʜ
    ʜ
    ʜ
    'MBUUFO
    
    
    
    
    

    View Slide

29. Dense Layer の⽋点
    • ⼊⼒のベクトルの全要素の相関をみている
    > 住宅価格予測みたいな話ならまだいい
    > もう作ってる特徴量と特徴量の組み合わせ
    - 例）東京墨⽥区 & 床⾯積 30m2 & 1K & ⾵呂トイレ別 & 新築 => 家賃⽉10万円
    • 「画像の特徴量」を抽出してからDense Layerに渡せば効率的
    !29
    ৞ࠐΈχϡʔϥϧωοτϫʔΫ $POWPMVUJPOBM/FVSBM/FUXPSL$//
    
    

    View Slide

30. 畳み込みニューラルネットワークの基礎
    !30
    

    View Slide

31. 畳み込み (Convolution)
    CURRENT_PIXEL_VALUE = 82
    NEW_PIXEL_VALUE =
    (-1 * 144) + (0 * 60) + (-2 * 19)
    + (0.5 * 188) + (4.5 * 82) + (-1.5 * 32)
    + (1.5 * 156) + (2 * 55) + (-3 * 27)
    !31
    
    
    
    
    
    
    u
    ijm
    =
    K 1
    X
    k=0
    W 1
    X
    p=0
    H 1
    X
    q=0
    z(l 1)
    i+p,j+q,k
    h
    pqkm
    + b
    ijm
    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
    https://news.yahoo.co.jp/byline/yuasamakoto/20190403-00120722/
    

    View Slide

32. !32
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    https://www.bbkong.net/fs/alleyoop/molten_BGL7
    

    View Slide

33. stride=(1, 1), padding=ʻvalidʼ
    !33
    

    View Slide

34. stride=(1, 1), padding=ʻvalidʼ
    !34
    

    View Slide

35. stride=(1, 1), padding=ʻvalidʼ
    !35
    

    View Slide

36. stride=(1, 1), padding=ʻvalidʼ
    !36
    

    View Slide

37. stride=(1, 1), padding=ʻvalidʼ
    !37
    

    View Slide

38. stride=(1, 1), padding=ʻvalidʼ
    !38
    

    View Slide

39. stride=(1, 1), padding=ʻvalidʼ
    !39
    

    View Slide

40. stride=(2, 2), padding=ʻvalidʼ
    !40
    

    View Slide

41. stride=(2, 2), padding=ʻvalidʼ
    !41
    

    View Slide

42. stride=(2, 2), padding=ʻvalidʼ
    !42
    

    View Slide

43. stride=(1, 1), padding=ʻsameʼ
    !43
    

    View Slide

44. stride=(1, 1), padding=ʻsameʼ
    !44
    

    View Slide

45. stride=(1, 1), padding=ʻsameʼ
    !45
    

    View Slide

46. stride=(1, 1), padding=ʻsameʼ
    !46
    

    View Slide

47. Max Pooling
    !47
    https://www.coursera.org/learn/introduction-tensorflow
    

    View Slide

48. AlexNet
    !48
    

    View Slide

49. !49
    TMJEFGSPN,FZOPUF4QFFDICZ-BVSFODF.PSPOFZBU%FFQ-FBSOJOH%BZ
    

    View Slide

50. !50
    https://medium.com/@lmoroney_40129/codelabs-from-googlemlsummit-f9d53cac8d24
    

    View Slide

51. !51
    https://medium.com/@lmoroney_40129/codelabs-from-googlemlsummit-f9d53cac8d24
    

    View Slide

52. 応⽤的なCNNアーキテクチャの紹介
    !52
    

    View Slide

53. ImageNet コンペの優勝モデル
    !53
    https://www.slideshare.net/ren4yu/ss-84282514
    

    View Slide

54. ResNet
    • 2015年のImageNetコンペ (ILSVRC) 優勝モデル
    • Residualモジュール（ショートカット機構）の導⼊
    !54
    http://image-net.org/challenges/talks/ilsvrc2015_deep_residual_learning_kaiminghe.pdf
    Revolution of Depth
    3.57
    6.7 7.3
    11.7
    16.4
    25.8
    28.2
    ILSVRC'15
    ResNet
    ILSVRC'14
    GoogleNet
    ILSVRC'14
    VGG
    ILSVRC'13 ILSVRC'12
    AlexNet
    ILSVRC'11 ILSVRC'10
    ImageNet Classification top-5 error (%)
    shallow
    8 layers
    19 layers
    22 layers
    152 layers
    Kaiming He, Xiangyu Zhang, Shaoqing Ren, & Jian Sun. “Deep Residual Learning for Image Recognition”. arXiv 2015.
    8 layers
    

    View Slide

55. !55
    Revolution of Depth
    ResNet, 152 layers
    1x1 conv, 64
    3x3 conv, 64
    1x1 conv, 256
    1x1 conv, 64
    3x3 conv, 64
    1x1 conv, 256
    1x1 conv, 64
    3x3 conv, 64
    1x1 conv, 256
    1x2 conv, 128, /2
    3x3 conv, 128
    1x1 conv, 512
    1x1 conv, 128
    3x3 conv, 128
    1x1 conv, 512
    1x1 conv, 128
    3x3 conv, 128
    1x1 conv, 512
    1x1 conv, 128
    3x3 conv, 128
    1x1 conv, 512
    1x1 conv, 128
    3x3 conv, 128
    1x1 conv, 512
    1x1 conv, 128
    3x3 conv, 128
    1x1 conv, 512
    1x1 conv, 128
    3x3 conv, 128
    1x1 conv, 512
    1x1 conv, 128
    3x3 conv, 128
    1x1 conv, 512
    1x1 conv, 256, /2
    3x3 conv, 256
    7x7 conv, 64, /2, pool/2
    Kaiming He, Xiangyu Zhang, Shaoqing Ren, & Jian Sun. “Deep Residual Learning for Image Recognition”. arXiv 2015.
    (there was an animation here)
    

    View Slide

56. !56
    Revolution of Depth
    ResNet, 152 layers
    1x1 conv, 512
    1x1 conv, 128
    3x3 conv, 128
    1x1 conv, 512
    1x1 conv, 256, /2
    3x3 conv, 256
    1x1 conv, 1024
    1x1 conv, 256
    3x3 conv, 256
    1x1 conv, 1024
    1x1 conv, 256
    3x3 conv, 256
    1x1 conv, 1024
    1x1 conv, 256
    3x3 conv, 256
    1x1 conv, 1024
    1x1 conv, 256
    3x3 conv, 256
    1x1 conv, 1024
    1x1 conv, 256
    3x3 conv, 256
    1x1 conv, 1024
    1x1 conv, 256
    3x3 conv, 256
    1x1 conv, 1024
    1x1 conv, 256
    3x3 conv, 256
    1x1 conv, 1024
    1x1 conv, 256
    3x3 conv, 256
    1x1 conv, 1024
    1x1 conv, 256
    3x3 conv, 256
    1x1 conv, 1024
    1x1 conv, 256
    3x3 conv, 256
    1x1 conv, 1024
    1x1 conv, 256
    Kaiming He, Xiangyu Zhang, Shaoqing Ren, & Jian Sun. “Deep Residual Learning for Image Recognition”. arXiv 2015.
    (there was an animation here)
    

    View Slide

57. !57
    Revolution of Depth
    ResNet, 152 layers
    1x1 conv, 256
    3x3 conv, 256
    1x1 conv, 1024
    1x1 conv, 256
    3x3 conv, 256
    1x1 conv, 1024
    1x1 conv, 256
    3x3 conv, 256
    1x1 conv, 1024
    1x1 conv, 256
    3x3 conv, 256
    1x1 conv, 1024
    1x1 conv, 256
    3x3 conv, 256
    1x1 conv, 1024
    1x1 conv, 256
    3x3 conv, 256
    1x1 conv, 1024
    1x1 conv, 256
    3x3 conv, 256
    1x1 conv, 1024
    1x1 conv, 256
    3x3 conv, 256
    1x1 conv, 1024
    1x1 conv, 256
    3x3 conv, 256
    1x1 conv, 1024
    1x1 conv, 256
    3x3 conv, 256
    1x1 conv, 1024
    1x1 conv, 256
    3x3 conv, 256
    1x1 conv, 1024
    1x1 conv, 256
    3x3 conv, 256
    1x1 conv, 1024
    1x1 conv, 256
    3x3 conv, 256
    Kaiming He, Xiangyu Zhang, Shaoqing Ren, & Jian Sun. “Deep Residual Learning for Image Recognition”. arXiv 2015.
    (there was an animation here)
    

    View Slide

58. !58
    Revolution of Depth
    ResNet, 152 layers
    3x3 conv, 256
    1x1 conv, 1024
    1x1 conv, 256
    3x3 conv, 256
    1x1 conv, 1024
    1x1 conv, 256
    3x3 conv, 256
    1x1 conv, 1024
    1x1 conv, 256
    3x3 conv, 256
    1x1 conv, 1024
    1x1 conv, 256
    3x3 conv, 256
    1x1 conv, 1024
    1x1 conv, 256
    3x3 conv, 256
    1x1 conv, 1024
    1x1 conv, 256
    3x3 conv, 256
    1x1 conv, 1024
    1x1 conv, 256
    3x3 conv, 256
    1x1 conv, 1024
    1x1 conv, 256
    3x3 conv, 256
    1x1 conv, 1024
    1x1 conv, 512, /2
    3x3 conv, 512
    1x1 conv, 2048
    1x1 conv, 512
    3x3 conv, 512
    1x1 conv, 2048
    1x1 conv, 512
    3x3 conv, 512
    1x1 conv, 2048
    ave pool, fc 1000
    Kaiming He, Xiangyu Zhang, Shaoqing Ren, & Jian Sun. “Deep Residual Learning for Image Recognition”. arXiv 2015.
    (there was an animation here)
    

    View Slide

59. ResNet
    !59
    Deep Residual Learning
    • Plaint net
    Kaiming He, Xiangyu Zhang, Shaoqing Ren, & Jian Sun. “Deep Residual Learning for Image Recognition”. arXiv 2015.
    any two
    stacked layers
    
    ()
    weight layer
    weight layer
    relu
    relu
    is any desired mapping,
    hope the 2 weight layers fit ()
    Deep Residual Learning
    • Residual net
    Kaiming He, Xiangyu Zhang, Shaoqing Ren, & Jian
    
    hop
    hop
    weight layer
    weight layer
    relu
    relu
    
    = +
    identity
    
    ()
    http://image-net.org/challenges/talks/ilsvrc2015_deep_residual_learning_kaiminghe.pdf
    

    View Slide

60. !60
    CIFAR-10 experiments
    0 1 2 3 4 5 6
    0
    5
    10
    20
    iter. (1e4)
    error (%)
    plain-20
    plain-32
    plain-44
    plain-56
    20-layer
    32-layer
    44-layer
    56-layer
    CIFAR-10 plain nets
    0 1 2 3 4 5 6
    0
    5
    10
    20
    iter. (1e4)
    error (%)
    ResNet-20
    ResNet-32
    ResNet-44
    ResNet-56
    ResNet-110
    CIFAR-10 ResNets
    56-layer
    44-layer
    32-layer
    20-layer
    110-layer
    • Deep ResNets can be trained without difficulties
    • Deeper ResNets have lower training error, and also lower test error
    solid: test
    dashed: train
    Kaiming He, Xiangyu Zhang, Shaoqing Ren, & Jian Sun. “Deep Residual Learning for Image Recognition”. arXiv 2015.
    

    View Slide

61. Further Reading
    • ⼀年半前 (2017-12) の資料ですが素晴らしいスライドです
    • 畳み込みニューラルネットワークの研究動向 by DeNA 内⽥さん
    • https://www.slideshare.net/ren4yu/ss-84282514
    !61
    

    View Slide

62. YOLO (You Only Look Once)
    !62
    https://www.youtube.com/watch?v=MPU2HistivI
    

    View Slide

63. YOLOのアーキテクチャ
    !63
    

    View Slide

64. Labeling / Annotation
    !64
    https://github.com/Microsoft/VoTT
    

    View Slide

65. AIモデルの運⽤・デプロイ
    !65
    

    View Slide

66. !66 https://www.programming-fun.net/article/article_40.html
    

    View Slide

67. !67
    https://www.youtube.com/watch?v=-UYyyeYJAoQ
    

    View Slide

68. モデルの利⽤
    !68
    O݄ؒϥϜμϊʔτ7PM /PΑΓ
    

    View Slide

69. モデルの利⽤
    !69
    O݄ؒϥϜμϊʔτ7PM /PΑΓ
    

    View Slide

70. !70
    

    View Slide

71. !71
    

    View Slide

72. !72
    

    View Slide

73. !73
    

    View Slide

74. !74
    

    View Slide

75. CNNモデルのデプロイ
    !75
    

    View Slide

76. !76
    

    View Slide

77. !77
    

    View Slide

78. !78
    curl --request POST \
    --url http://:9008/login \
    --header 'accept: application/json' \
    --header 'content-type: application/json' \
    --data '{"userId":"","password":""}'
    

    View Slide

79. Links
    • SKIL Download Page
    • https://skymind.ai/download
    • SKIL Documentation
    • https://docs.skymind.ai/docs
    • SKIL REST API reference
    • https://docs.skymind.ai/reference?newApiExplorer=true
    • SKIL Clients
    • https://github.com/SkymindIO/skil-clients
    !79
    

    View Slide

80. まとめ
    !80
    

    View Slide