Experimental study of Manifold learning and tangent propagation

Abstract

In many data mining problems dealing with artificially high-dimensional data, can cause a lot of difficulties. This is due to the fact that interpreting a high-dimensional data can be very challenging. There are several ways to avoid that kind of difficulties and one of them is the dimension reduction method. Nowadays, manifold learning approach is widely used in various dimension reduction problems. In this work we propose a new technique of finding data manifold by using auto-encoders.