Upgrade to Pro — share decks privately, control downloads, hide ads and more …

DevFest - Machine Learning on the Web

2b6d7bdd43058e87f53866eb86538a59?s=47 Galuh Sahid
November 30, 2019
Technology
2
190

DevFest - Machine Learning on the Web

DevFest SG 2019
GitHub: https://github.com/galuhsahid/ml-on-the-web

2b6d7bdd43058e87f53866eb86538a59?s=128

Galuh Sahid

November 30, 2019
Tweet

More Decks by Galuh Sahid

See All by Galuh Sahid
2b6d7bdd43058e87f53866eb86538a59?s=48 galuhsahid
0
64
2b6d7bdd43058e87f53866eb86538a59?s=48 galuhsahid
0
82
2b6d7bdd43058e87f53866eb86538a59?s=48 galuhsahid
1
180
2b6d7bdd43058e87f53866eb86538a59?s=48 galuhsahid
0
140
2b6d7bdd43058e87f53866eb86538a59?s=48 galuhsahid
0
130
2b6d7bdd43058e87f53866eb86538a59?s=48 galuhsahid
0
130
2b6d7bdd43058e87f53866eb86538a59?s=48 galuhsahid
0
59
2b6d7bdd43058e87f53866eb86538a59?s=48 galuhsahid
0
150
2b6d7bdd43058e87f53866eb86538a59?s=48 galuhsahid
0
45

Other Decks in Technology

See All in Technology
252e6c31a6452aa80deb9ad0107975c7?s=48 attakei
0
230
95aaab3d693889550fd2e7ae02f2f789?s=48 takaking22
1
1.5k
Eeac2f3b51b69b3cd6fb14ed348f0c6e?s=48 limes2018
0
180
Fb025419ed38f98df595eb2eafc859f4?s=48 ihcomega56
2
1.2k
2aea20b60d544d0dcccb70a4bb6af648?s=48 hminamiohtsu
3
1.3k
2a37329cdb03944fdf86b0e54397ff30?s=48 veliswaboya
1
110
97b3cca999b52cb5296675ac0a5c12cd?s=48 emiki
2
230
A5fed8c15d016c2a42d7d9f830326045?s=48 takuros
2
1.5k
842515eaf8fbb2dfcc75197e7797dc15?s=48 sat
1
540
6e28f76cf4ddc493b1df1d284e0d3400?s=48 hhj4ck
0
2.5k
1741a7877b6ff93bf29af4d11c2c895e?s=48 ryan403
0
150
285e58eafc0cffa03bdb9f612768434c?s=48 kelvinksps
1
130

Featured

See All Featured
44b54d2ffcb7857004071cc6795dde11?s=48 cassininazir
349
20k
C44e1f7e22c3f23cff7bc130871047ef?s=48 eileencodes
117
26k
5b9fe87ec1faa67a4599782930f45ec9?s=48 sstephenson
149
12k
A9704266587836f7e784235e5073b93e?s=48 jmmastey
19
4.9k
E4f5d494ebdc9e7cce1aecf3ce3e8bc1?s=48 shpigford
372
43k
053e75a5b48b44d6dd0612795dfb326d?s=48 notwaldorf
48
2.9k
B3d6434763caa0ef5dc4b792662c49f7?s=48 smashingmag
287
47k
1177e050db6bafe62885362edf6e3537?s=48 lauravandoore
17
2k
D8fc2580cfaca035f666d9e4ee79a7f7?s=48 mongodb
24
4k
E837d2996f52008ace055a08fbfff992?s=48 frogandcode
129
20k
C4de88ea47538e4594750e3861a341c8?s=48 brettharned
96
3.2k
6bb82b13cda86d3b821fa44100335ddf?s=48 thoeni
6
780

Transcript

  1. None

  2. Machine Learning on the Web DevFest SG - November 30,

    2019

  3. Hi! I’m Galuh Twitter: @galuhsahid https://galuh.me

  4. Machine learning gives computers the ability to learn without being

    explicitly programmed. - Arthur Samuel (1959)

  5. Computer Rules Output Data

  6. Computer Data Output Rules Computer Rules Output Data

  7. Ingredients of machine learning

  8. Data: Pictures https://www.rainforest-alliance.org/sites/default/files/styles/750w_585h/public/2016-09/three-toed-sloth.jpg?itok=uWF-NdZZ https://www.thoughtco.com/thmb/DpVtY--UXVOtG68m9LhMY7be2w0=/768x0/filters:no_upscale():max_bytes(150000):strip_icc()/happy-red-panda-171399380-5b574325c9e77c005b690b41.jpg

  9. Data: Voice

  10. Data: Text

  11. Features

  12. Features

  13. Features

  14. Mathematical representation of a real life process. Model 2000*number of

    floors + … + 5000*number of rooms = house price

  15. 2000*number of floors + … + 5000*number of rooms =

    house price Model error = $5400

  16. 2000*number of floors + … + 5000*number of rooms =

    house price Model error = $5400 3000*number of floors + … + 5000*number of rooms = house price error = $4000

  17. 2000*number of floors + … + 5000*number of rooms =

    house price Model error = $5400 3000*number of floors + … + 5000*number of rooms = house price error = $4000 3000*number of floors + … + 2000*number of rooms = house price error = $2500

  18. Machine learning on the web?

  19. Machine learning on the web? “A WebGL accelerated JavaScript library

    for training and deploying ML models.”

  20. Why TensorFlow.js? Browser: !Without installation

  21. Why TensorFlow.js? Browser: !Without installation !Interactive

  22. Augmented Reality Speech Recognition Accessibility Education Conversational AI

  23. Why TensorFlow.js? Browser: !Without installation !Interactive !Data stays in the

    client

  24. Why TensorFlow.js? Browser: !Without installation !Interactive !Data stays in the

    client Server: !Nicely integrates with existing node.js stack

  25. Why TensorFlow.js? Browser: !Without installation !Interactive !Data stays in the

    client Server: !Nicely integrates with existing node.js stack

  26. How come?

  27. None

  28. Machine learning on the web? A machine learning library for

    JavaScript built on top of TensorFlow.js

  29. Machine learning on the web? A machine learning library for

    JavaScript built on top of TensorFlow.js

  30. Running pre-trained models

  31. Image classification app

  32. Input: file upload

  33. MobileNet Mobile-first computer vision model Small, low-latency, low-power

  34. MobileNet Mobile-first computer vision model Small, low-latency, low-power

  35. Image classification with MobileNet MobileNet Prediction: Bicycle (0.7845)

  36. None
  37. None
  38. None

  39. https://teachablemachine.withgoogle.com/

  40. https://teachablemachine.withgoogle.com/

  41. https://teachablemachine.withgoogle.com/

  42. None
  43. None

  44. Input: webcam

  45. None
  46. None

  47. MobileNet

  48. MobileNet DarkNet

  49. Retraining existing models

  50. New data MobileNet Prediction: ? Prediction: ?

  51. Should we train from scratch? 14 million images

  52. Transfer learning MobileNet trained on a new task Prediction: Angklung

    (0.873)

  53. Rock Paper Scissors

  54. Rock Paper Scissors: Demo

  55. None
  56. None
  57. None
  58. None
  59. None
  60. None
  61. None

  62. Transfer learning MobileNet Bicycle (0.373) Cup (0.232) Vacuum (0.121)

  63. Transfer learning MobileNet Bicycle (0.373) Cup (0.232) Vacuum (0.121) x

  64. Transfer learning MobileNet Rock (0.923) Paper (0.422) Scissors (0.231)

  65. Transfer learning Credit: MathWorks

  66. None
  67. None
  68. None
  69. None
  70. None

  71. Scroll with Your Voice

  72. Scroll with Your Voice: Demo

  73. Sound Classification: Data

  74. Sound Classification: Model

  75. Sound Classification: Model CNN Prediction: Up (0.86)

  76. None
  77. None
  78. None

  79. … and many more https://ml5js.org/reference/

  80. Building new models from scratch

  81. index.html <html> <head> <script src="https://cdn.jsdelivr.net/npm/@tensorflow/tfjs/dist/ tf.min.js"></script> </head> <body> <div id="output_field"></div>

    </body> … Credit: Laurence Moroney

  82. index.html … <script> async function learnLinear(){ const model = tf.sequential();

    model.add(tf.layers.dense({units: 1, inputShape: [1]})); } learnLinear(); </script> Credit: Laurence Moroney

  83. index.html … <script> async function learnLinear(){ const model = tf.sequential();

    model.add(tf.layers.dense({units: 1, inputShape: [1]})); model.compile({ loss: 'meanSquaredError', optimizer: 'sgd' }); } learnLinear(); </script> Credit: Laurence Moroney

  84. index.html … <script> async function learnLinear(){ const model = tf.sequential();

    model.add(tf.layers.dense({units: 1, inputShape: [1]})); model.compile({ loss: 'meanSquaredError', optimizer: 'sgd' }); const xs = tf.tensor2d([-1, 0, 1, 2, 3, 4], [6, 1]); const ys = tf.tensor2d([-3, -1, 1, 3, 5, 7], [6, 1]); } learnLinear(); </script> Credit: Laurence Moroney

  85. index.html … <script> async function learnLinear(){ const model = tf.sequential();

    model.add(tf.layers.dense({units: 1, inputShape: [1]})); model.compile({ loss: 'meanSquaredError', optimizer: 'sgd' }); const xs = tf.tensor2d([-1, 0, 1, 2, 3, 4], [6, 1]); const ys = tf.tensor2d([-3, -1, 1, 3, 5, 7], [6, 1]); } learnLinear(); </script> Credit: Laurence Moroney y=2x-1

  86. index.html … <script> async function learnLinear(){ const model = tf.sequential();

    model.add(tf.layers.dense({units: 1, inputShape: [1]})); model.compile({ loss: 'meanSquaredError', optimizer: 'sgd' }); const xs = tf.tensor2d([-1, 0, 1, 2, 3, 4], [6, 1]); const ys = tf.tensor2d([-3, -1, 1, 3, 5, 7], [6, 1]); await model.fit(xs, ys, { epochs: 500 }); } learnLinear(); Credit: Laurence Moroney

  87. index.html … model.compile({ loss: 'meanSquaredError', optimizer: 'sgd' }); const xs

    = tf.tensor2d([-1, 0, 1, 2, 3, 4], [6, 1]); const ys = tf.tensor2d([-3, -1, 1, 3, 5, 7], [6, 1]); document.getElementById("output_field").innerText = model.predict( tf.tensor2d([10], [1, 1]) ); } learnLinear(); </script> </html> Credit: Laurence Moroney y=2x-1 2(10)-1 = 19
  88. None

  89. What’s next? ! A Beginner’s Guide to Machine Learning in

    JavaScript by Daniel Shiffman ! Magenta.js - music & art with machine learning ! Coursera: Machine Learning by Andrew Ng ! Machine learning + hardware ! ml5.js source code ! Bias in machine learning - book reference: Weapons of Math Destruction oleh Cathy O’Neil

  90. Thank you!