- TensorFlow 1.x Deep Learning Cookbook
- Antonio Gulli Amita Kapoor
- 32字
- 2021-07-02 22:01:39
Getting ready
For function approximation, the loss function should be the MSE. The inputs should be normalized and, while the hidden layer can be ReLU, the output layer should preferably be sigmoid.