Papers
7
Total Citations
41
H-Index
4
About
Alexander Gepperth is a leading researcher in continual learning, robotics, and multimodal perception, whose work bridges the gap between machine learning and autonomous systems. His most impactful contribution is the development of PROPRE, a neural learning paradigm that autonomously extracts meaningful concepts from multimodal data flows by leveraging predictability across sensory modalities—a breakthrough for unsupervised, incremental learning in embodied agents. Gepperth’s research on efficient online bootstrapping of sensory representations (10 citations) and neural network fusion of color, depth, and location for object instance recognition (14 citations) has advanced real-time robotic perception, enabling mobile robots to learn from synthetic data and adapt without costly manual labeling. Notably, his 2022 empirical study on continual learning methods for Q-learning (6 citations) pioneers the integration of CL into reinforcement learning, addressing non-stationary data distributions—a critical challenge for lifelong learning agents. With over 40 citations across his top works, Gepperth’s contributions to simultaneous concept formation and incremental object classification have shaped developmental learning in robotics, making him a key figure in creating adaptive, self-improving AI systems.
Research Focus
Key Achievements
Top Papers
- 1
- 2Efficient online bootstrapping of sensory representations10 citations · 2012
- 3A Study of Continual Learning Methods for Q-Learning6 citations · 2022
- 4Incremental learning for bootstrapping object classifier models4 citations · 2016
- 5Simultaneous concept formation driven by predictability3 citations · 2012
- 6
- 7