Deep Learning in Time Series Analysis

7 Clustering for Learning at Deep Level

The concept of deep learning is focused on neural networks with many layers and neurons, even up to thousands, containing several hyperparameters for the training, needing heuristic and sometimes naturally inspired methods for the optimization. This can be certainly true to a certain extent, but definitely cannot cover the entire scope of deep learning. This chapter provides another view towards deep learning based on using unsupervised methods to help the learning process for extracting subtle contents of time series at the deep level. This chapter focuses on the potentials of the well-known clustering methods in the deep learning perspective, that can strengthen the learning process.

Segregation of data ...