I am Deep Learning Freelance. I want to share my last project of Churn prediction with variational auto-encoder. Auto-encoders are a great example of unsupervised learning, and there a lot we can do with them. Unbalanced classification problems, can be analyzed and solved with auto-encoders. One of the greatest challenges in supervised Deep Learning project is data labeling. Problem gets much worse in segmentation problems, when we need to label each pixel in a large data set. In unsupervised learning, there are no labeling, that's one of the reasons why unsupervised learning is so interesting.
