Perform sentiment analysis with LSTMs, using TensorFlow

In order to understand how deep learning can be applied, think about all the different forms of data that are used as inputs into machine learning or deep learning models. Convolutional neural networks use arrays of pixel values, logistic regression uses quantifiable features, and reinforcement learning models use reward signals. The common theme is that the inputs need to be scalar values, or matrices of scalar values. When you think of NLP tasks, however, a data pipeline like this may come to mind. This kind of pipeline is problematic.