I am Adrián Rodríguez-Muñoz, a 3rd year grad student at MIT EECS under the supervision of Prof. Antonio Torralba. My research focuses on learning how to use all data effectively, such as low-quality and out-of-distribution data in generative models, and even non-realistic procedurally generated data in vision models. Previously, I worked on efficient adversarial robustness. I am currently doing an internship at Amazon Research, working on retrieval augmented generation (RAG) with diffusion models.

Prior to my PhD, I studied Mathematics and Data Science and Engineering at CFIS-UPC until 2022. During my undergrad, I did internships at the Barcelona Supercomputing Center, the biotech company ZeClinics, and the finance firms Aspect Capital (London) and Susquehanna International Group (Dublin).

Click here for my CV.

Publications

Ambient Diffusion Omni: Training Good Models with Bad Data

Adrián Rodríguez-Muñoz*, Giannis Daras*, Adam Klivans, Antonio Torralba, Constantinos Daskalakis

[paper] [webpage] [code]

[ICML 2025] Separating Knowledge and Perception with Procedural Data

Adrián Rodríguez-Muñoz, Manel Baradad, Phillip Isola, Antonio Torralba

[paper] [webpage] [code]

[ECCV 2024] Characterizing Model Robustness with Natural Input Gradients

Adrián Rodríguez-Muñoz, Tongzhou Wang, Antonio Torralba

[paper] [webpage] [code]

[CVPR 2024 - Highlight] A Vision Check-up for Language Models

Pratyusha Sharma*, Tamar Rott Shaham*, Manel Baradad, Stephanie Fu, Adrián Rodríguez-Muñoz, Shivam Duggal, Phillip Isola, Antonio Torralba

[paper] [webpage]

Aliasing is a Driver of Adversarial Attacks

Adrián Rodríguez-Muñoz, Antonio Torralba

[paper] [webpage] [code]