Summary

In this chapter, we looked at improving the performance of a neural network. Although we worked with a lightweight dataset, we have learned some important ideas around improving our model’s performance–ideas that will come in handy, both in the exam and on the job. You now know that data quality and model complexity are two sides of the machine learning coin. If you have good-quality data, a poor model will yield subpar results and, on the flip side, even the most advanced model will yield a suboptimal result with bad data.

By now, you should have a good understanding and hands-on experience of fine-tuning neural networks. Like a seasoned expert, you should be able to understand the art of fine-tuning hyperparameters and apply this to different machine learning problems and not just image classification. Also, you have seen that model building requires a lot of experimenting. There is no silver bullet, but having a good understanding of the moving parts and...