Cristian Meo

Hi! My name is Cristian Meo, I’m a final-year PhD student at Delft University of Technology, supervised by Justin Dauwels and Geert Leus, and mentored by Anirudh Goyal.

I’m currently working on Generative Models for video prediction and model based RL downstream tasks. My current research focus is leveraging pretrained Vision-Language Models (VLMs) to learn Generalist World Models.

Along my academic career I obtained a BSc in Mechanical Engineering at Politecnico di Torino and a MSc in BioRobotics at TUDelft cum Laude. I have also spent six months as a research intern at Mila in Montreal, supervised by Yoshua Bengio and Anirudh Goyal.

I am in the job market right now. To get in touch, write me an email at: c.meo@tudelft.nl

News

Dec 15, 2024	Excited to share two incredible milestones from this year! Our paper “Masked Generative Priors Improve World Models Sequence Modelling Capabilities” received the Outstanding Paper Award at the World Models workshop at ICLR 2025. Additionally, our team secured a substantial Snellius Compute Grant of 31.9M GPU SBUs (~€400,000) on the national supercomputer to further advance our research in world models. Grateful for the amazing collaborators and looking forward to pushing the state-of-the-art even further!
Jun 01, 2024	Excited and grateful to be among a hand-picked group of students and researchers heading to San Francisco this June for the YC AI Startup School! Looking forward to an incredible learning experience and the opportunity to connect with fellow innovators in the AI space.
May 20, 2024	I am thrilled to announce that our paper, “Extreme Precipitation Nowcasting using Transformer-based Generative Models,” has been published as a SPOTLIGHT paper at the ICLR 2024 Workshop on Tackling Climate Change with Machine Learning! You can find the full announcement here and the preprint on ArXiv.
May 19, 2024	I am extremely happy to announce that our paper “Alpha-TCVAE: On the Relationship Between Disentanglement and Diversity” has been accepted at ICLR2024! Find the Full Paper Here. This research was conducted during my internship at Mila - Quebec Artificial Intelligence Institute, under the amazing supervision of Prof. Yoshua Bengio and Anirudh Goyal, Research Scientist at Google DeepMind.

Selected Publications

AlphaTC-VAE: On the Relationship Between Disentanglement and Diversity

Cristian Meo, Louis Mahon, Anirudh Goyal, and 1 more author

The Twelfth International Conference on Learning Representations (ICLR2024), 2024

Bib HTML

@article{meo2024alphatcvae,
  title = {AlphaTC-VAE: On the Relationship Between Disentanglement and Diversity},
  author = {Meo, Cristian and Mahon, Louis and Goyal, Anirudh and Dauwels, Justin},
  journal = {The Twelfth International Conference on Learning Representations (ICLR2024)},
  year = {2024},
}

Extreme Precipitation Nowcasting using Transformer-based Generative Models

Cristian Meo, Ankush Roy, Mircea Lica, and 6 more authors

ICLR 2024 workshop on Tackling Climate Change with Machine Learning, 2024

Spotlight Bib HTML

true

@article{meo2024extreme,
  title = {Extreme Precipitation Nowcasting using Transformer-based Generative Models},
  author = {Meo, Cristian and Roy, Ankush and Lica, Mircea and Yin, Junzhe and Che, Zeineb Bou and Wang, Yanbo and Imhoff, Ruben and Uijlenhoet, Remko and Dauwels, Justin},
  journal = {ICLR 2024 workshop on Tackling Climate Change with Machine Learning},
  year = {2024}
}

Object-Centric Temporal Consistency via Conditional Autoregressive Inductive Biases

Cristian Meo, Akihiro Nakano, Aniket Didolkar, and 6 more authors

NeurIPS 2024 Workshop on Compositional Learning: Perspectives, Methods, and Paths Forward., 2024

Bib HTML

@article{meo2024casa,
  title = {Object-Centric Temporal Consistency via Conditional Autoregressive Inductive Biases},
  author = {Meo, Cristian and Nakano, Akihiro and Didolkar, Aniket and Suzuki, Masahiro and Goyal, Anirudh and Zhang, Mengmi and Dauwels, Justin and Matsuo, Yutaka and Bengio, Yoshua},
  journal = {NeurIPS 2024 Workshop on Compositional Learning: Perspectives, Methods, and Paths Forward.},
  year = {2024}
}

Masked Generative Priors Improve World Models Sequence Modelling Capabilities

Cristian Meo, Mircea Lica, Zarif Ikram, and 6 more authors

ICLR 2025 workshop on World Models: Understanding, Modelling, and Scaling., 2025

Outstanding Paper Award Bib HTML

true

@article{meo2025git-storm,
  title = {Masked Generative Priors Improve World Models Sequence Modelling Capabilities},
  author = {Meo, Cristian and Lica, Mircea and Ikram, Zarif and Nakano, Akihiro and Shah, Vedant and Didolkar, Aniket and Liu, Dianbo and Goyal, Anirudh and Dauwels, Justin},
  journal = {ICLR 2025 workshop on World Models: Understanding, Modelling, and Scaling.},
  year = {2025}
}