Reproduce “Auto.gov: Learning-based Governance for Decentralized Finance (DeFi)” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Jiahua Xu, Yebo Feng, Daniel Perez, Benjamin Livshits
The Auto.gov framework introduces a Deep Q-Network reinforcement learning agent for automated Decentralized Finance (DeFi) governance, enabling rapid parameter adjustments that enhance attack resilience and profitability. It increased protocol profit by over 60% against price oracle attacks in simulations and achieved a tenfold increase in profitability compared to a static baseline using real-world data.
Key insights
- A learning-based governance agent can make strategic, infrequent parameter adjustments, recognizing that excessive changes can negatively impact user confidence and protocol revenue.
- The agent learns to adapt its governance strategy, becoming more conservative in the presence of attack risks while maximizing profit in stable conditions.
- Deep Q-Networks are crucial for managing the complex, continuous state spaces of DeFi governance, outperforming simpler tabular Q-learning methods.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.