Abstract: Graphs are a powerful framework for modeling complex systems, such as social, biological, communication, and infrastructure networks. Generative models for graphs have a long history in network science, starting with random graph models in the 50s. In the last few decades, network science has led to several advancements toward generating graphs that reproduce properties seen in the real-world (e.g. degree distributions, clustering). However, network science models, such as Preferential Attachment, are able to generate only the graph topology and their limited number of parameters lacked enough flexibility to capture more than a handful of properties. More recently, deep generative models for graphs have achieved promising results, learning graph models directly from data by extending ideas from computer vision to graph domains. The most successful case studies for these models have been for molecular graphs and, to a lesser extent, code generation. Still, graph generative models failed to achieve the same success as their vision and language counterparts. In this talk, we will discuss some of the key challenges for graph generative models and how modern results from language and vision, such as transformers and diffusion, can help us in addressing these challenges. In particular, we will use physical graphs (e.g. mesh discretizations) and cybersecurity to motivate new research directions on graph generative models.
Bio: My research focuses on developing algorithms and models for mining and learning from complex datasets, broadly defined as data science, especially for data represented as graphs/networks.
I’m particularly interested in problems motivated by computational social science, infrastructure, and healthcare. The tools that I apply to address these problems include machine learning, network science, graph theory, linear algebra, optimization, and statistics.
I got a Ph.D in Computer Science from the University of California, Santa Barbara, advised by Ambuj Singh, where I was also a postdoctoral scholar. Before that, I got a B.Sc and M.Sc degrees in Computer Science from Universidade Federal de Minas Gerais, in Brazil, advised by Wagner Meira Jr. I’ve also been a visiting scholar at the Rensselaer Polytechnic Institute, hosted by Mohammed J. Zaki.