Deep learning can say a set of AI (AI) machine learning networks that can learn from unstructured or unlabeled knowledge. This uses multiple layers to remove collections at higher levels from raw inputs and heaps. As an example, lower layers in image technique can ensure edges, whereas higher layers can ensure that ideas are important to someone like digits or letters or faces. Deep Learning is associated AI performing that imitates human brain processing in process information to be used in higher cognitive processes. Deep Learning AI is capable of discovering from information that each is unstructured and unlabeled. Deep learning, a range of machine learning, can make sight fraud or concealment simpler. This paper mainly focuses on the ideas of Deep Learning, why we should we use Deep Learning over Machine Learning, its basic architectures, characteristics and the limitation. The main intention of this paper is to explore and present a comprehensive survey of Deep Learning awareness among technical people, deep learning applications, architectures used, and any contribution of the various applications worldwide at intervals. The paper ends with the conclusion and future aspects of Deep Learning.

Keywords : Artificial Intelligence, Machine Learning, Neural Networks.