This is my master thesis project. The goal of this project is to optimize the existing implementation of the machine learning model Sum-Product Network in order to make faster and efficient learning.
Project Details
The project consist in a first part of study the literature about this new ML model. After several studies the python library deeprob-kit has been decided to improve.
The implementation consist on adding some alternative methods to apply Structural Learning algorithm. With this algorithm the model can be built through the analysis of the input data instead of the manual drawing of the researcher.
The optimizations have been tested on several database in order to validate the implementation.