This chapter is dedicated to helping you understand more of the Microsoft Cognitive Toolkit, or CNTK. The inspiration for the examples contained within this chapter comes from the Python version of CNTK 106: Part A – Time Series prediction with LSTM (Basics). As C# developers, the Python code is not what we will be using (although there are several ways in which we could) so we made our own C# example to mirror that tutorial. To make our example easy and intuitive, we will use the Sine function to predict future time-series data. Specifically, and more concretely, we will be using a long short-term memory recurrent neural network, sometimes called an LSTM-RNN or just LSTM. There are many variants of the LSTM; we will be working with the original.

In this chapter, we will cover the following topics:

• LSTM
• Tensors
• Static and dynamic axis
• Loading datasets
• Plotting data
• Creating models
• Creating mini-batches
• And more…