如何使用Keras构建一个resnet来训练和预测主类中的子类？

Question

I would like go implement a hierarchical resnet architecture. However, I could not find any solution for this. For example, my data structure is like:

• class A
  • Subclass 1
  • Subclass 2
  • ....
• class B
  • subclass 6
  • ........

So i would like to train and predict the main class and then the subclass of the chosen/predicted mainclass. Can someone provide a simple example how to do this with generators?

Answer 1

The easiest way to do so would be to train multiple classifiers and build a hierarchical system by yourself.

One classifier detecting class A, B etc. After that make a new prediction for subclasses.

If you want only one single classifier: What about just killing the first hierarchy of parent classes? Should be also quite easy. If you really want a model, where the hierarchy is learned take a look at Hierarchical Multi-Label Classification Networks.