Discounted UCB
An exploration strategy that introduces a discount factor into UCB so that the weight of earlier observations decays over time, adapting to non-stationary environments in which reward distributions change.
Timeline (last 90 days)
HEMO uses a discounted UCB scheduler to direct adaptation resources toward more promising modules based on changes in each module's utility.
Discounted UCB introduces a discount factor on top of UCB, causing the weight of early observations to decay over time, thereby adapting to non-stationary environments where the reward distribution changes.
All Facts (2)
HEMO uses a discounted UCB scheduler to direct adaptation resources toward more promising modules based on changes in each module's utility.
50%UnverifiedDiscounted UCB introduces a discount factor on top of UCB, causing the weight of early observations to decay over time, thereby adapting to non-stationary environments where the reward distribution changes.
50%