MSc.
Jorge da Silva Gonçalves
PhD Student
- jorge.dasilvagoncalves@inf.ethz.ch
- Address
-
Department of Computer Science
CAB G 15.2
Universitätstr. 6
CH – 8092 Zurich, Switzerland - Room
- CAB G 15.2
I completed my Bachelor’s degree in Banking and Finance and with a minor Statistics at the University of Zurich, focusing on quantitative methods and classical statistics. During this time, I worked for several years at the Chair of Mathematics for Business and Economics under Prof. Dr. Christiane Barz and wrote my thesis on tree-based models under the supervision of Prof. Dr. Michael Wolf.
I then obtained a Master’s in Statistics at ETH Zurich, where I focused more on machine learning. My Master’s thesis at the Medical Data Science group explored enhancing the generative capabilities of a generative clustering model, involving hierarchical VAEs and diffusion models. In May 2024, I joined the group as a research intern and now continue as a doctoral student.
My research interests include probabilistic and generative AI, clustering, and interpretable machine learning, with applications in healthcare.
Publications
This paper introduces Diffuse-TreeVAE, a deep generative model that integrates hierarchical clustering into the framework of Denoising Diffusion Probabilistic Models (DDPMs). The proposed approach generates new images by sampling from a root embedding of a learned latent tree VAE-based structure, it then propagates through hierarchical paths, and utilizes a second-stage DDPM to refine and generate distinct, high-quality images for each data cluster. The result is a model that not only improves image clarity but also ensures that the generated samples are representative of their respective clusters, addressing the limitations of previous VAE-based methods and advancing the state of clustering-based generative modeling.
AuthorsJorge da Silva Gonçalves, Laura Manduchi, Moritz Vandenhirtz, Julia E. Vogt
SubmittedICML 2024 Workshop on Structured Probabilistic Inference & Generative Modeling
Date27.07.2024