Francesco Locatello
Papers
3
Total Citations
37
H-Index
3
About
Francesco Locatello is a leading researcher at the intersection of representation learning and reinforcement learning, with a primary focus on disentanglement, out-of-distribution (OOD) generalization, and the transfer of inductive biases. His major contributions include pioneering work on understanding how pretrained representations can enable sample-efficient agents to generalize in real-world settings, a critical step toward achieving higher-level cognition in AI. Locatello is perhaps best known for his influential research on disentanglement, where he introduced a new benchmark dataset to study the transfer of inductive biases from simulation to reality—a foundational challenge in representation learning. His most cited work, “On the Transfer of Inductive Bias from Simulation to the Real World: a New Disentanglement Dataset” (2019, 30 citations), has become a key reference for researchers aiming to learn compact, semantically meaningful representations without relying on costly real-world data. Locatello’s ongoing work on OOD generalization in reinforcement learning continues to shape how the field approaches building robust, generalizable agents, making him a notable voice in modern AI research.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3