Dimensionality Reduction using Deep Learning Techniques
1B. ChandraSekhar, Abboju Niranjan, Dr G.VenkataRamiReddy
Dimensionality reduction is a compelling way to deal with scaling back the information. It is a strategy that endeavors to extend a lot of high dimensional vectors to a lower dimensionality space while holding measurements among them. The Artificial Intelligence and information mining systems may not be successful for high-dimensional information due to the scourge of dimensionality and question precision and efficiency will debase quickly as the measurement increments. It is a typical preprocessing step for feature extraction, classification and different undertakings. Learning a classifier on low-dimensional data sources is quick. All the more critically, Dimensionality reduction can help to get familiar with a superior classifier, especially when the information has a low-dimensional structure, and with little datasets, where it has a regularizing impact that can help abstain from over fitting. This paper deals with the working of deep learning technique for dimensionality reduction on various datasets. It is also compared with the traditional machine learning method i.e. Principal Component Analysis (PCA).
Autoencoder, Dimensionality Reduction, Principal Component Analysis (PCA), Reconstruction of data.