TENSORFLOW

Introduction to tensorflow

Transcript

2. What is Neural Network ? Little brief in deep learning ? Intro to Tensorflow Demo: Detecting handwriting digits using MNIST Sotoy Spam Detection sample using Neural Network
3. There is 2 way of machine can learn : 1. Using Statistical Learning (SVM, Logistic Regression, Linear Regression, NaiveBayes, Decision Tree) 2. Using Neural Network
4. Basically neural network is a statistical learning But, the different is Neural Network worked based on How Our Brain Works
5. Credits to Sungjoon (https://github.com/sjchoi86/dl_tutorials)
6. Credits to Sungjoon (https://github.com/sjchoi86/dl_tutorials)
7. BASIC SINGLE LAYER NETWORK Input: 1 2 4 0 W: 1 3 −2 1 0 0 −2 1 1 2 0 0 Bias: 2 1 3 −1 −8 5 1 −7 8 1 0 8 Credits to Sungjoon (https://github.com/sjchoi86/dl_tutorials)
8. (NEURAL) NETWORK Credits to Sungjoon (https://github.com/sjchoi86/dl_tutorials)
9. ARTIFICIAL NEURAL NETWORK Input Layer Output Layer Hidden Layers Credits to Sungjoon (https://github.com/sjchoi86/dl_tutorials)
10. MULTI-LAYER PERCEPTRON Credits to Sungjoon (https://github.com/sjchoi86/dl_tutorials)
11. ACTIVATION FUNCTION Credits to Sungjoon (https://github.com/sjchoi86/dl_tutorials)
12. COST FUNCTION This is just a function that we want to minimize. There is no guarantee that it will bring us to the best solution! It should be differentiable! Credits to Sungjoon (https://github.com/sjchoi86/dl_tutorials)
13. What is Neural Network ? Little brief in deep learning ? Intro to Tensorflow Demo: Detecting handwriting digits using MNIST Sotoy Spam Detection sample using Neural Network Credits to Sungjoon (https://github.com/sjchoi86/dl_tutorials)
14.  Basically Deep Learning = Neural Network  But the different is the layer is deep ( > 1 )  So just add the layer ? Many people think of that  Between every layer there is some technique to calculate the move of the data  Like our brain works
15. What is Neural Network ? Little brief in deep learning ? Intro to Tensorflow Demo: Detecting handwriting digits using MNIST Sotoy Spam Detection sample using Neural Network
16. TENSORFLOW 16 Tensor + Flow Tensor: n-dimensional arrays Vector: 1-dimensional tensor Matrix: 2-dimensional tensor Deep learning processes are flows of tensors A sequence of tensor operations Can also represent a number of machine learning algorithms Credits to Sungjoon (https://github.com/sjchoi86/dl_tutorials)
17. RELU (RECTIFIER LINEAR UNIT) NETWORK 17 Inputs Outputs Credits to Sungjoon (https://github.com/sjchoi86/dl_tutorials)
18. MATRIX OPERATION 18 Credits to Sungjoon (https://github.com/sjchoi86/dl_tutorials)
19. IN TENSORFLOW 19 Credits to Sungjoon (https://github.com/sjchoi86/dl_tutorials)
20. VARIABLES 20 Credits to Sungjoon (https://github.com/sjchoi86/dl_tutorials)
21. TENSORFLOW 21 Credits to Sungjoon (https://github.com/sjchoi86/dl_tutorials)
22. SESSION 22 Credits to Sungjoon (https://github.com/sjchoi86/dl_tutorials)
23. INITIALIZE VARIABLE 23 Credits to Sungjoon (https://github.com/sjchoi86/dl_tutorials)
24. PLACEHOLDER 24 Credits to Sungjoon (https://github.com/sjchoi86/dl_tutorials)
25. FEED 25 Credits to Sungjoon (https://github.com/sjchoi86/dl_tutorials)
26. PREDICTION 26 Credits to Sungjoon (https://github.com/sjchoi86/dl_tutorials)
27. ERROR 27 Credits to Sungjoon (https://github.com/sjchoi86/dl_tutorials)
28. LOSS 28 Credits to Sungjoon (https://github.com/sjchoi86/dl_tutorials)
29. OPTIMIZATION 29 Credits to Sungjoon (https://github.com/sjchoi86/dl_tutorials)
30. ITERATIVE UPDATE 30 Credits to Sungjoon (https://github.com/sjchoi86/dl_tutorials)
31. BIASES 31 Credits to Sungjoon (https://github.com/sjchoi86/dl_tutorials)
32. GO DEEPER 32 Credits to Sungjoon (https://github.com/sjchoi86/dl_tutorials)
33. What is Neural Network ? Little brief in deep learning ? Intro to Tensorflow Demo: Detecting handwriting digits using MNIST Sotoy Spam Detection sample using Neural Network
34. What is Neural Network ? Little brief in deep learning ? Intro to Tensorflow Demo: Detecting handwriting digits using MNIST Sotoy Spam Detection sample using Neural Network