Now, let's take a look how the DL4J provides us with these functionalities through the transfer learning API it has. The DL4J transfer learning API enables users to (see more at https://deeplearning4j.org/transfer-learning):
- Modify the architecture of an existing model
- Fine-tune learning configurations of an existing model
- Hold parameters of a specified layer (also called a frozen layer) constant during training
These functionalities are depicted in the following diagram, where we solve task B (similar to task A) using the transfer learning technique:
Working principle of transfer learning
In the next section, ...
