KEYWORD |
Visualization of deep learning models
Thesis in external company
Reference persons FABRIZIO LAMBERTI
Research Groups GR-09 - GRAphics and INtelligent Systems - GRAINS
Description The goal of this thesis is to implement user-friendly, interactive representations of deep learning models from multi-dimensional data. Deep learning are black-box models with million of parameters. Effective visualization can help users understand how the network is behaving. The goal of this thesis is to design re-usable tools that can be integrated in a variety of applications. Such tools could be used either by an expert user (to simplify the training process), or by a non expert user, to understand network behaviour. Examples such visualization tools are similarity layouts such as t-SNE (t-distributed stochastic neighbor embedding), visualization of network activations, etc.. Skills possessed or to acquire: programming skills (Python, C++ or Java, Keras/Tensorflow or other deep learning framework). This thesis could be carried out in collaboration with a company.
See also http://grains.polito.it/work.php
Deadline 13/02/2019
PROPONI LA TUA CANDIDATURA