Contextual Bandits

Contextual bandits extend the multi-armed bandit framework by incorporating observable contextual information before each decision, allowing agents to learn action-context relationships and make more informed choices. Rather than blindly selecting arms, the agent observes features about the current situation and uses learned policies to map contexts to optimal actions. This sophisticated approach better reflects real-world decision-making where actions should adapt based on circumstances, significantly improving decision quality and convergence speed compared to context-free bandits. Example: Aave's dynamic interest rate mechanism implicitly uses contextual bandit logic by adjusting borrowing rates based on contextual factors like market conditions, collateral types, and protocol utilization rates, optimizing capital efficiency within the lending market. Why it matters for AI and data in Web3: Contextual bandits enable DeFi protocols to dynamically adjust fees, rates, and mechanisms based on observable market conditions and user characteristics, creating responsive systems that optimize outcomes for specific market contexts rather than applying one-size-fits-all strategies.

Explore the full Web3 Glossary — 2,062+ expert-curated definitions. Need guidance? Talk to our consultants.