Summary
In this chapter, we implemented a deep CNN and achieved an excellent recognition result exceeding 99% for handwritten digit recognition. We also discussed the motivation for making a network deeper and the current tendency toward deeper networks. We also looked at the trends and applications of deep learning, and the research is accelerating it, which will advance this technology into the future.
In the field of deep learning, there is much that is still unknown, and new research is being published all the time. Researchers and engineers around the world continue to research actively and will realize technologies that we cannot even imagine yet.
The following points were covered in this chapter:
- Making a network deeper will improve performance for many deep learning problems.
- In image recognition competitions, techniques using deep learning get a high ranking, and current networks are deeper than their predecessors
- Famous networks include VGG, GoogLeNet...
