Scholar
Antoine Dedieu
Google Scholar ID: Hgoc3FUAAAAJ
Senior Research Scientist at Google DeepMind
Citations & Impact
Citations
798
H-index
11
i10-index
12
Publications
20
Co-authors
13
list available
Publications
Resume (English only)
Academic Achievements
- “Improving Transformer World Models for Data-Efficient RL” received Outstanding Paper Award at ICLR 2025 World Model Workshop; to appear at ICML 2025.
- Co-first author of “DMC-VB: A Benchmark for Representation Learning for Control with Visual Distractors,” NeurIPS 2024.
- Contributed to “Diffusion Model Predictive Control,” TMLR 2025.
- “Learning Cognitive Maps from Transformer Representations for Efficient Planning in Partially Observed Environments,” ICML 2024.
- “Schema-learning and rebinding as mechanisms of in-context learning and emergence,” NeurIPS 2023 Spotlight.
- “Learning noisy-OR Bayesian Networks with Max-Product Belief Propagation,” ICML 2023.
- “Perturb-and-max-product: Sampling and learning in discrete energy-based models,” NeurIPS 2021.
- “Sample-Efficient L0-L2 Constrained Structure Learning of Sparse Ising Models,” AAAI 2021.
Background
- Senior Research Scientist at Google DeepMind.
- Currently interested in building agents that can efficiently learn new tasks in new environments.
- Believes such agents must learn a latent world model combining (a) a representation model mapping observations to a compact latent space, and (b) a generative world model describing latent dynamics.
- Thinks LLMs/VLMs can serve as rich priors for solving new tasks.
- Research lies at the intersection of representation learning, model-based reinforcement learning, and in-context learning.
