ISI Programming Course - 07 - Tensorflow

Transcript

1. Jungwon Seo | University of Stavanger ISI Programming Class Tensorflow
2. Tensorflow • Originally developed by Google. • Mainly for Deep Neural Network.
 
 
 
 
 

3. Why do we need to use this?
4. Flexibility, Scalability, Portability
5. Again! Neural Network
6. Neuron
7. Activation Function
8. Backpropagation
9. Gradient descent
10. Minimum Error
11. Learning Rate
12. 1. Convolutional Neural Network • Good fit with image recognition.
13. 2. Recurrent Neural Network • Good fit with text and audio data.
14. Deep learning can’t solve the problem that you don’t know how to solve.