Summary

Transfer learning has gained traction in the deep learning community, due to its improved performance, speed, and accuracy in building deep learning models. We discussed the rationale behind transfer learning and explored transfer learning as a feature extractor and a fine-tuned model. We built a couple of solutions using the top-performing pre-trained models and saw how they outperformed our baseline model when applied to the X-ray dataset.

By now, you should have gained a solid understanding of transfer learning and its applications. Equipped with this knowledge, you should be able to apply transfer learning as either a feature extractor or a fine-tuned model when building real-world deep learning solutions for a wide range of tasks.

With this, we have come to the end of this chapter and this section of the book. In the next chapter, we will discuss natural language processing (NLP), where we will build exciting NLP applications using TensorFlow.