Solving Sequence Problems with LSTM in Keras: Part 2
This is the second and final part of the two-part series of articles on solving sequence problems with LSTMs. In the part 1 of the series, I explained how to solve one-to-one and many-to-one sequence problems using LSTM. In this part, you will see how to solve one-to-many and many-to-many sequence problems via LSTM in Keras.
Image captioning is a classic example of one-to-many sequence problems where you have a single image as input and you have to predict the image description in the form of a word sequence. Similarly, stock market prediction for the next X days, where input is the stock price of the previous Y days, is a classic example of many-to-many sequence problems.
In this article you will see very basic examples of one-to-many and many-to-many problems. However, the concepts learned in this article will lay the foundation for solving advanced sequence problems, such as stock price prediction and automated image captioning that we will see in the upcoming articles.
One-to-Many Sequence Problems
One-to-many sequence problems are the type of sequence problems where input data has one time-step and the output contains a vector of multiple values or multiple time-steps. In this