TensorFlow is an open source software library for numerical computation using data flow graphs.
내용은 모두를 위한 딥러닝 시즌2 강의영상을 참고하였습니다.
https://www.youtube.com/watch?v=OR_NwgouflE&list=PLQ28Nx3M4Jrguyuwg4xe9d9t2XE639e5C&index=36
* Process Flow
1. Set hyperparameters - learning rate, training epochs, batch size, etc.
2. Make a data-pipeline - use tf.data
3. Build a neural network model - use tf.keras sequential API
4. Define a loss function - cross entropy
5. Calculate gradient - use tf.GradientTape
6. Select an optimizer - Adam optimizer
7. Define a metric for model's performance - accuracy
8. (optional) Make a checkpoint for saving
9. Train and Validate a NN model.
'딥러닝' 카테고리의 다른 글
| Transfer Learning (전이학습) (0) | 2021.08.17 |
|---|---|
| CNN 기초 (0) | 2021.05.25 |
| Introduction to Deep Learning : 딥러닝의 시작 (0) | 2021.01.02 |