An Analysis on Deep Learning with its Advancements

Most recently, Deep Learning being a part of wider area of Machine Learning plays a vital role of study from data
representations. DL has been rapidly growing in several application domains by means of its different approaches,
methods and tools. It is designed with a network of interconnected neuron units resembling the structure and
function of bio neuron. Deep learning network has the ability of studying unsupervised from data collection which
is not structured or labeled, supervised data containing labels, semi-supervised data which is partially supervised
also known as Reinforcement data. Deep learning upgrades the concepts of machine learning to the next successive
level. Deep learning algorithms has been constructed with multilayered connections, which makes use of artificial
neural networks to study multiple levels related to the different levels of abstraction to solve complicated problems.
Expected Outcomes have been shown that deep learning accurates the learning data features compared to traditional
machine learning concepts in the various sectors. Deep learning architectures are Deep Neural Networks, Deep Belief
Networks, Convolutional Neural Networks, AutoEncoders and so on. Every interconnected layer represents a depth
of knowledge. This paper confers a brief study on Deep learning, its categories and its advanced features such as
AutoEncoders, Generative Adversarial Networks and its types , Multi-view learning and Multi-task learning.