BIG DATA WITH DEEP LEARNING (SUPERVISED VS. UNSUPERVISED): A SURVEY

摘要

Due to the dramatic increment in technology development in the last few years, a huge volume of data is being produced every day for each person. This data is coming from different types of applications (like google search, online shopping history, etc) and it is called Big Data. Big Data represents a massive stream of data which is continuously growing and changing, and it is needed to be controlled, analyzed and monetized to help the decision makers of organizations to innovate their companies. In addition to that, this analysis impacts the existing and future technology. Artificial Intelligence and Machine Learning algorithms are being adopted to provide effective automated tools and operations on big data (like data labeling, analyzing, diagnosing and so). One of the techniques that can be used by machine learning is the "neural networks", and in this case it is called as Deep Learning. Deep learning could be unsupervised learning which is concerned with labeling and segmenting these large amounts of various unlabeled full of noise information. Or, it could be supervised learning which is concerned with recognizing the labeled data to recognize patterns in it, and then to be translated into valuable insights for further implementation. This paper is a review that survey the supervised and unsupervised deep learning techniques, clarify when and how to be used, and give an overview on the technologies and applications which are adopting them.