Who's Afraid Of Machine Learning? & first steps with TensorFlow

Slide 1

Slide 1 text

Who’s afraid of Machine Learning? Britt Barak

Slide 2

Slide 2 text

Britt Barak Google Developer Expert - Android Women Techmakers Israel Britt Barak @brittBarak

Slide 3

Slide 3 text

No content

Slide 4

Slide 4 text

No content

Slide 5

Slide 5 text

No content

Slide 6

Slide 6 text

No content

Slide 7

Slide 7 text

No content

Slide 8

Slide 8 text

No content

Slide 9

Slide 9 text

No content

Slide 10

Slide 10 text

No content

Slide 11

Slide 11 text

In a machine...

Slide 12

Slide 12 text

No content

Slide 13

Slide 13 text

Strawberry Not Strawberry

Slide 14

Slide 14 text

Input Red Seeds pattern Top leaves 0.64 0.75 0.4

Slide 15

Slide 15 text

0.64 0.75 0.4 Input Red Seeds pattern Top leaves

Slide 16

Slide 16 text

0.64 0.75 0.4 Input Red Seeds pattern Top leaves

Slide 17

Slide 17 text

0.64 0.75 0.4 Input Red Seeds pattern Top leaves

Slide 18

Slide 18 text

0.64 0.75 0.4 Input 0.5 0.8 0.3 Red Seeds pattern Top leaves

Slide 19

Slide 19 text

0.64 0.75 0.4 Input Red Seeds pattern Top leaves 0.5 0.8 0.3 0.5 * 0.64 + 0.8 * 0.75 + 0.3 * 0.4

Slide 20

Slide 20 text

0.64 0.75 0.4 Input Red Seeds pattern Top leaves 0.5 0.8 0.3 0.5 * 0.64 + 0.8 * 0.75 + 0.3 * 0.4 ___________ 1.04

Slide 21

Slide 21 text

0.64 0.75 0.4 Input Red Seeds pattern Top leaves 0.5 0.8 0.3 0.5 * 0.64 + 0.8 * 0.75 + 0.3 * 0.4 ___________ 1.04 + 0.7

Slide 22

Slide 22 text

0.64 0.75 0.4 1.74 0.5 * 0.64 + 0.8 * 0.75 + 0.3 * 0.4 ___________ 1.04 + 0.7 ___________ 1.74 Input Red Seeds pattern Top leaves 0.5 0.8 0.3

Slide 23

Slide 23 text

0.64 0.75 0.4 1.02 1.74 Input Red Seeds pattern Top leaves 0.97

Slide 24

Slide 24 text

0.64 0.75 0.4 Input Red Seeds pattern Top leaves 1.02 1.74 0.97

Slide 25

Slide 25 text

0.64 0.75 0.4 Output Strawberry Not Strawberry Input Red Seeds pattern Top leaves 1.02 1.74 0.97 0.87 0.13

Slide 26

Slide 26 text

0.64 0.75 0.4 0.87 0.13 Strawberry Not Strawberry Output Input Red Seeds pattern Top leaves 1.02 1.74 0.97

Slide 27

Slide 27 text

No content

Slide 28

Slide 28 text

0.7 0.03 0.01 3.72 0.89 1.92 Strawberry Not Strawberry Output Input Red Seeds pattern Top leaves 0.2 0.8

Slide 29

Slide 29 text

0.7 0.03 0.01 3.72 0.89 1.92 Strawberry Not Strawberry Output Input Red Seeds pattern Top leaves 0.2 0.8

Slide 30

Slide 30 text

0.7 0.03 0.01 3.72 0.89 1.92 0.2 0.8 Strawberry Not Strawberry Output Input Red Seeds pattern Top leaves

Slide 31

Slide 31 text

0.5 * 0.64 + 0.8 * 0.75 + 0.3 * 0.4 ___________ 1.04 + 0.7 ___________ 1.74 Strawberry Not Not Strawberry Not Not Strawberry Not Not

Slide 32

Slide 32 text

Training TRAINING

Slide 33

Slide 33 text

0.64 0.75 0.4 1.02 1.74 0.97 0.89 0.11 Strawberry Not Strawberry Output Input Red Seeds pattern Top leaves

Slide 34

Slide 34 text

Strawberry Not Strawberry Output Input Hidden Red Seeds pattern Top leaves

Slide 35

Slide 35 text

No content

Slide 36

Slide 36 text

Data science

Slide 37

Slide 37 text

We get a trained model !

Slide 38

Slide 38 text

TensorFlow - Open source - Widely used - Flexible for scale: - 1 or more CPUs / GPUs - desktop, server, mobile device

Slide 39

Slide 39 text

Strawberry

Slide 40

Slide 40 text

Strawberry

Slide 41

Slide 41 text

Strawberry ● Bandwidth ● Performance ● Latency ● Network ● Security ● Privacy ● …

Slide 42

Slide 42 text

TensorFlow Mobile - Speech Recognition - Image Recognition - Object Localization - Gesture Recognition - Translation - Text Classification - Voice Synthesis

Slide 43

Slide 43 text

Lightweight Fast Cross platform

Slide 44

Slide 44 text

MobileNet Inception-V3 SmartReply Models

Slide 45

Slide 45 text

No content

Slide 46

Slide 46 text

Image Classifier classifier .classify(bitmap) label

Slide 47

Slide 47 text

1. Add Assets

Slide 48

Slide 48 text

No content

Slide 49

Slide 49 text

labels.txt strawberry orange lemon fig pineapple banana jackfruit custard apple pomegranate hay carbonara chocolate sauce dough meat loaf

Slide 50

Slide 50 text

2. Add TensorFlow Lite

Slide 51

Slide 51 text

repositories {  maven {  url 'https://google.bintray.com/tensorflow'  }  }    dependencies {  // ...  implementation 'org.tensorflow:tensorflow-lite:+'  } build.gradle

Slide 52

Slide 52 text

android {  aaptOptions {  noCompress "tflite"  }  } build.gradle

Slide 53

Slide 53 text

3. Create ImageClassifier.java

Slide 54

Slide 54 text

Image Classifier

Slide 55

Slide 55 text

ImageClassifier.java model = loadModelFile();  tflite = new Interpreter();

Slide 56

Slide 56 text

ImageClassifier.java model = loadModelFile();  tflite = new Interpreter(model);

Slide 57

Slide 57 text

MappedByteBuffer loadModelFile() {  AssetFileDescriptor descriptor= getAssets().openFd(MODEL_PATH);  

Slide 58

Slide 58 text

Slide 59

Slide 59 text

MappedByteBuffer loadModelFile() {  AssetFileDescriptor descriptor= getAssets().openFd(MODEL_PATH);  FileInputStream inputStream = new FileInputStream(descriptor.getFileDescriptor()); FileChannel channel = inputStream.getChannel(); long start = descriptor.getStartOffset();  long length = descriptor.getDeclaredLength();  return channel.map(FileChannel.MapMode.READ_ONLY, start, length); }

Slide 60

Slide 60 text

Image Classifier [strawberry, apple, ... ] labels.txt

Slide 61

Slide 61 text

ImageClassifier.java model = loadModelFile();  tflite = new Interpreter(model); labelList = loadLabelList();

Slide 62

Slide 62 text

labels.txt strawberry orange lemon fig pineapple banana jackfruit custard apple pomegranate hay carbonara chocolate sauce dough meat loaf

Slide 63

Slide 63 text

List loadLabelList() throws IOException {  InputStreamReader inputStream = new InputStreamReader(getAssets().open(LABEL_PATH));   }

Slide 64

Slide 64 text

List loadLabelList() throws IOException {  InputStreamReader inputStream = new InputStreamReader(getAssets().open(LABEL_PATH)); BufferedReader reader = new BufferedReader(inputStream);    }

Slide 65

Slide 65 text

List loadLabelList() throws IOException {  InputStreamReader inputStream = new InputStreamReader(getAssets().open(LABEL_PATH)); BufferedReader reader = new BufferedReader(inputStream);  List labelList = new ArrayList<>(); String line;  while ((line = reader.readLine()) != null) {  labelList.add(line);  }    }

Slide 66

Slide 66 text

Slide 67

Slide 67 text

Image Classifier [ [0..6] , [ 0.1 ] , ... ] [strawberry, apple, ... ] probArray labels.txt

Slide 68

Slide 68 text

probArray = { [0.7], [0.3], [0], [0], } labelList = { strawberry, apple, pineapple, banana, } 0.3

Slide 69

Slide 69 text

ImageClassifier.java model = loadModelFile();  tflite = new Interpreter(model); labelList = loadLabelList(); probArray = new float[1][labelList.size()];

Slide 70

Slide 70 text

Image Classifier [......] [ [0..6] , [ 0.1 ] , ... ] [strawberry, apple, ... ] ByteBuffer probArray labels.txt

Slide 71

Slide 71 text

ImageClassifier.java model = loadModelFile();  tflite = new Interpreter(model); labelList = loadLabelList(); probArray = new float[1][labelList.size()]; imgData =  ByteBuffer.allocateDirect( DIM_IMG_SIZE_X * DIM_IMG_SIZE_Y * DIM_PIXEL_SIZE); imgData.order(ByteOrder.nativeOrder());

Slide 72

Slide 72 text

4. Run the model / classify

Slide 73

Slide 73 text

classifier .classify(bitmap) Image Classifier [......] [ [0..6] , [ 0.1 ] , ... ] [strawberry, apple, ... ] ByteBuffer probArray labels.txt

Slide 74

Slide 74 text

ImageClassifier.java String classify(Bitmap bitmap) {    convertBitmapToByteBuffer(imgData, bitmap);      }

Slide 75

Slide 75 text

void convertBitmapToByteBuffer(Bitmap bitmap) { //... bitmap.getPixels(intValues, 0, bitmap.getWidth(), 0, 0,bitmap.getWidth(), bitmap.getHeight()); }

Slide 76

Slide 76 text

void convertBitmapToByteBuffer(Bitmap bitmap) { //... bitmap.getPixels(intValues, 0, bitmap.getWidth(), 0, 0,bitmap.getWidth(), bitmap.getHeight()); int pixel = 0; for (int i = 0; i < DIM_IMG_SIZE_X; ++i) { for (int j = 0; j < DIM_IMG_SIZE_Y; ++j) { final int val = intValues[pixel++]; imgData.put((byte) ((val >> 16) & 0xFF)); imgData.put((byte) ((val >> 8) & 0xFF)); imgData.put((byte) (val & 0xFF)); } } }

Slide 77

Slide 77 text

Slide 78

Slide 78 text

ImageClassifier.java String classify(Bitmap bitmap) {    convertBitmapToByteBuffer(imgData, bitmap);    tflite.run(imgData, probArray);      }

Slide 79

Slide 79 text

ImageClassifier.java String classify(Bitmap bitmap) {    convertBitmapToByteBuffer(imgData, bitmap);    tflite.run(imgData, probArray);    String textToShow = getTopLabels();  return textToShow;  }

Slide 80

Slide 80 text

Strawberry - 0.87 Apple - 0.13 Tomato - 0.01

Slide 81

Slide 81 text

Machine Learning is a new world

Slide 82

Slide 82 text

Links - Tensorflow - https://www.tensorflow.org/ - Tensorflow lite - https://www.tensorflow.org/mobile/tflite/ - Codes labs - codelabs.developers.google.com/codelabs/tensorflow-for-poets-2-tflite/ - Google’s Machine Learning Crash Course - developers.google.com/machine-learning/crash-course/ - [Dr. Joe Dispenza]

Slide 83

Slide 83 text

Thank you! Keep in touch! Britt Barak @brittBarak