Speaker: Filipi Nascimento Silva

Abstract:
Visualization is crucial for understanding intricate systems and complex machine learning pipelines. It helps users gain insights, interpret results more easily, and supports the process of hypothesis generation. Helios-Web is a novel web-based platform designed for visualizing large networks and embeddings in real-time, even with millions of nodes or points. Accessible through any web browser, Helios-Web has a wide range of applications. We will demonstrate different methodologies to explore and interactively visualize large embeddings. To tackle scalability and avoid cluttered visualizations, we use advanced computer graphics techniques and GPU power, enhancing interactivity. All these features are implemented in Helios-Web, which we use to showcase visualizations in three key areas: a) the Science of Science field, where it presents publications, authors, and other scholarly data in interactive 2D and 3D maps; b) integration with the OSoMe platform to visualize social media networks and embeddings; and c) its use in biological networks. These new tools and methods represent a significant advancement in our ability to visualize, understand, and interact with the complex systems underlying large-scale data