Borislav Mavrin
Papers
1
Total Citations
4
H-Index
1
About
Borislav Mavrin is a researcher whose work lies at the intersection of distributional reinforcement learning and efficient exploration. His most notable contribution is the Quantile Option Architecture (QUOTA), introduced in a 2018 paper that has garnered 4 citations. QUOTA represents a significant conceptual shift in RL: rather than relying solely on the mean of a value distribution for decision-making, Mavrin’s approach leverages quantiles, providing a richer, more nuanced signal for action selection. This innovation opens a new dimension for exploration, allowing agents to consider tail risks and opportunities that traditional methods might overlook. By grounding exploration in the structure of the value distribution itself, Mavrin’s work offers a principled alternative to heuristic exploration strategies. His research is particularly relevant for domains where uncertainty is high and robust decision-making is critical. Though early in its citation trajectory, QUOTA has already influenced subsequent work in distributional RL and option discovery, marking Mavrin as a thoughtful contributor to the ongoing evolution of reinforcement learning algorithms.
Research Focus
Key Achievements
Top Papers
- 1QUOTA: The Quantile Option Architecture for Reinforcement Learning4 citations · 2018