Publication: Entropy Triangles: An application to measuring the transmission of entropy in Autoencoders
Loading...
Identifiers
Publication date
2019-10
Defense date
2019-10-14
Authors
Advisors
Tutors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
The aim of this Bachelor’s Thesis is to test the functionalities of the Entropy Trian-
gle. The Entropy Triangle is a tool designed to analyse the fow of information on a
variable after it has been processed by means of a transformation, a classifcation, or any
other type method that may have shaped its informational characteristics. The appropri-
ate environment has been created with the objective to obtain an accurate profle of such
properties, selecting three diferent datasets with a wide variety of traits that allow for
our tool to display its more interesting features. The experiments performed include an
in-depth analysis of their results and its impacts on the data science feld.
Description
Keywords
Deep Neural Networks, Information Bottleneck, Autoencoder, Entropy Triangle, Classifcation, Data Science