Types of unsupervised DL
There are broadly three different types of UL methods that are currently available:
- Clustering
- Anomaly detection
- Association
Let’s discuss each of these in detail in the following sections.
Clustering
Clustering, as the name suggests, is a type of UL method to group similar data points in the training dataset—for example, the clustering of tissues based on the gene expression values from genomics data. This is the most common method of UL. Here, the DL models look for similar data points in the training data to group them using the appropriate distance measurement method (Figure 7.1). One challenge with the clustering method is you need to predefine the number of clusters for the algorithm to group clustering based on the number of clusters. However, there are methods out there that can help arrive at this cluster size to input into the learning algorithm.
Figure 7.1 – Clustering of multiple...