What can AutoML do?
- Automatically search neural network model structure
- Get better results through Fine-tuning/migration learning on specified tasks
How to search the network structure?
- Search must have search space
- There must be feedback during the search process
- Automate the search process
Search space ---- Take the convolutional neural network as an example,
only the convolutional layer is retained, and the last layer uses global pooling instead of each convolutional layer. There are five parameters
Filter height/width/number
Stride height/width
N-layer convolutional nerve The network has N*5 parameter
search feedback, which
indicates the quality of the neural network
. The accuracy of the model on a certain data set can be used. The model
needs to be trained