Volker Tresp

Papers

Total Citations

116

H-Index

About

Volker Tresp is a prominent researcher whose work sits at the intersection of reinforcement learning, goal-conditioned learning, and experience replay optimization. His most influential contributions address a fundamental challenge in multi-goal reinforcement learning: how agents can efficiently learn from imbalanced or poorly prioritized experience data. Tresp's research has systematically advanced the field of hindsight experience replay and trajectory prioritization. His 2019 paper on Maximum Entropy-Regularized Multi-Goal Reinforcement Learning (47 citations) introduced principled approaches to balancing goal-conditioned policy learning, while his contemporaneous work on Curiosity-Driven Experience Prioritization (42 citations) tackled the critical problem of imbalanced goal-state distributions using density estimation techniques. His earlier Energy-Based Hindsight Experience Prioritization (27 citations) laid important groundwork by moving beyond random experience replay toward value-aware selection strategies. Collectively, these contributions represent a coherent research agenda: making reinforcement learning agents smarter about *what* they learn from, not just *how* they learn. With over 116 citations across just three papers, Tresp's work has earned meaningful traction in the RL community, particularly among researchers working on robotic manipulation, sparse-reward environments, and sample-efficient learning systems.