L3S Best Publication Award Q3+Q4/2025
Category: Green AI
Hybrid pathfinding optimization for the Lightning Network with Reinforcement Learning
Authors: D. Valko, D. Kudenko
Published in Engineering Applications of Artificial Intelligence
The paper in a nutshell:
The Bitcoin Lightning Network enables fast, low-cost transactions, but payments sometimes fail because the system struggles to find reliable routes across the network. In our new paper, we combine reinforcement learning (RL) with existing routing algorithms to dynamically choose better paths for payments. Testing on a real network snapshot shows that this hybrid approach can increase payment success rates by up to ~10% in challenging network conditions, while also maintaining efficiency and reducing unnecessary long-distance routing that can increase infrastructure load.
These results highlight how RL can help improve the reliability and scalability of next-generation digital payment systems without changing their core architecture.
Link to the paper: https://www.sciencedirect.com/science/article/abs/pii/S0952197625002258
