Conceptvariance-reduced estimation / 方差缩减

Variance Reduction Estimation

A technique in stochastic optimization that reduces the variance of stochastic estimates by maintaining reference gradients or reference estimates. Representative methods include SVRG and SARAH; in reinforcement learning, the technique is used to stabilize value function updates.

Timeline (last 90 days)

Oct 7

The algorithm employs variance-reduced estimation of the Bellman optimality operator locally at each agent.

Unverified50%

All Facts (1)

Unverified

The algorithm employs variance-reduced estimation of the Bellman optimality operator locally at each agent.

50%

Source Articles

Adversarially Robust Federated Q-Learning: Learning Optimal Policies from Unreliable TrajectoriesOct 7