Scholar
Christian Gumbsch
Google Scholar ID: FQOFw5cAAAAJ
Postdoc, University of Amsterdam
Citations & Impact
Citations
283
H-index
8
i10-index
7
Publications
20
Co-authors
11
list available
Publications
Resume (English only)
Academic Achievements
- ICML 2025: 'SENSEI - Semantic Exploration Guided by Foundation Models For Learning Versatile World Models' (*equal contribution).
- ICLR 2024: 'THICK - Learning Hierarchical World Models with Adaptive Temporal Abstractions' (Spotlight, top 5%).
- IEEE ICDL 2022: 'Developing Hierarchical Anticipations from Event Segmentation' (SmartBot Challenge winner, oral presentation).
- Nature Machine Intelligence 2022: 'Intelligent Problem-Solving as Integrated Hierarchical Reinforcement Learning'.
- NeurIPS 2021: 'GateL0RD - Sparsely Changing Latent States for Prediction and Planning in POMDPs' (26% acceptance rate).
- IEEE TCDS 2019: 'Segmenting Behavioral Primitives from Sensorimotor Exploration for Event-Based Planning'.
Background
- Fascinated by how humans and other animals learn adaptive goal-directed behavior from experience.
- Aims to develop autonomous embodied agents with similar capabilities.
- Two main research directions:
- - AI research: Improving decision-making in deep learning agents through exploration, long-horizon planning, and knowledge reuse, focusing on generalization, temporal abstraction, world models, and intrinsic motivation.
- - Cognitive modeling: Building computational models to understand human adaptive behavior mechanisms, including event segmentation, epistemic gaze behavior, and multimodal speech processing.
