In this chapter, we explored a TensorFlow-trained neural network using TensorBoard and trained our own modified version of that network with different epochs and learning rates. This gave you hands-on experiences on how to train a highly performant neural network and also allowed you to explore some of its limitations.
Do you think we can achieve similar accuracy with real Bitcoin data? We will attempt to predict future Bitcoin prices using a common neural network algorithm during Chapter 5, Model Architecture. In Chapter 6, Model Evaluation and Optimization, we will evaluate and improve that model and, finally, in Chapter 7, Productization, we will create a program that serves the prediction of that system via an HTTP API.
