The rise of the Internet of Things, data storage and management models has led to an explosion in the quantity and complexity of the data we collect today. To understand these data and better manipulate them, experts are relying on information visualization. However, the actual complexity of these new data makes former visualizations ineffective. It becomes necessary to design new visualization techniques and review the evaluation methods used to measure their effectiveness.
This thesis presents contributions to the field of Information Visualization on two main aspects: automatic generation and evaluation of visualizations. For these two axes, our work leverages machine and deep learning techniques that have demonstrated their ability to efficiently process large volumes of data. The applications of the contributions presented in this manuscript make use of Machine and Deep Learning models for Graph Drawing with Node-Link representations, Overlap Removal in point clouds (scatter plots), and Automated Evaluation of Visualizations.