Let’s continue our discussion on use cases of transfer learning using some examples from natural language processing.
Let’s revisit the example of document summarisation. If you want to do document summarisation in some other language, such as Hindi, you can take the following steps:
- Use word embeddings in English to train a document summarisation model (assuming a significant amount of data in English is available)
- Use word embeddings of another language such as Hindi (where you have a data crunch) to tune the English summarisation model
Let’s now summarise the main idea of transfer learning.
In the next segment, you will see some common use cases of transfer learning applied to computer vision tasks.
Report an error