Summary:
The following paper’s main goal is to study the up-and-coming, machine learning
field of Deep Learning. The main focus was to analyze and study the following
interrelated chapters of the field:
• Analysis of the learning methods, supervised, unsupervised and hybrid
learning, used while training deep learning models. Emphasis was laid on
presenting the Back-Propagation algorithm, the cornerstone of most of the
available training methods. Emphasis was also placed on presenting the
difficulties that may occur while training a deep learning model, as well as
some of the available solutions.
• Presentation and analysis of many of the available models (neural
networks) used in modern deep learning applications
• Presentation of the main features of the available deep learning
software suites.
Lastly, for the practical section of the study, there has been a construction
of deep learning models, with the use of python as the chosen programming
language, as well as with the use of some of the software solutions presented
in the theoretical part.
