Constrained Reinforcement Learning
A framework that introduces cost constraints beyond the standard reinforcement learning objective of maximizing cumulative reward, typically formalized as optimizing reward subject to expected cost remaining below a threshold. It is suited to safety-critical scenarios.
Timeline (last 90 days)
Constrained reinforcement learning introduces constraints on certain costs in addition to maximizing cumulative reward, typically formalized as optimizing reward subject to the expected cost being below a threshold.
Compared to adjusting reward weights, constrained reinforcement learning offers two advantages: communicability and composability, as adding a new constraint does not force the recalibration of all weights.
All Facts (2)
Constrained reinforcement learning introduces constraints on certain costs in addition to maximizing cumulative reward, typically formalized as optimizing reward subject to the expected cost being below a threshold.
50%UnverifiedCompared to adjusting reward weights, constrained reinforcement learning offers two advantages: communicability and composability, as adding a new constraint does not force the recalibration of all weights.
50%