Georg Martius

Full Professor@Eberhard-Karls-Universität Tübingen·德国·OpenReview

研究方向

model-based reinforcement learning · representation learning · minimal description length · reinforcement learning · intrinsic motivation · information Theory · dynamics systems · self-organization · robotics

Georg Martius

Test-time Offline Reinforcement Learning on Goal-related Experience

Differentiable Simulation of Hard Contacts with Soft Gradients for Learning and Control

Learning 3D-Gaussian Simulators from RGB Videos

Revive and Recouple: Mitigating Plasticity Loss in Transformer Architectures

Forecasting in Offline Reinforcement Learning for Non-stationary Environments

SENSEI: Semantic Exploration Guided by Foundation Models to Learn Versatile World Models

Zero-Shot Offline Imitation Learning via Optimal Transport

Active Fine-Tuning of Generalist Policies

Active Fine-Tuning of Multi-Task Policies

SENSEI: Semantic Exploration Guided by Foundation Models to Learn Versatile World Models

Zero-Shot Offline Imitation Learning via Optimal Transport

Advancing Out-of-Distribution Detection via Local Neuroplasticity

On the Transfer of Object-Centric Representation Learning

Offline vs. Online Learning in Model-based RL: Lessons for Data Collection Strategies