Conditional Geometric Resampling
A technique for estimating unobserved losses that estimates the expected loss through resampling from a geometric distribution, controlling variance while preserving unbiasedness.
Timeline (last 90 days)
The loss estimator designed in the paper combines three key components: forced exploration, conditional geometric resampling, and an adaptive learning rate with surrogate probabilities.
Conditional geometric resampling estimates the expected loss of unobserved nodes by resampling from a geometric distribution, ensures unbiasedness of the estimate conditional on the given history, and controls variance through the tail properties of the geometric distribution.
Through the design of conditional geometric resampling and surrogate probabilities, the algorithm requires only a constant number of graph operations per update, thereby achieving nearly linear expected time without explicitly enumerating all unobserved nodes.
All Facts (3)
The loss estimator designed in the paper combines three key components: forced exploration, conditional geometric resampling, and an adaptive learning rate with surrogate probabilities.
50%UnverifiedConditional geometric resampling estimates the expected loss of unobserved nodes by resampling from a geometric distribution, ensures unbiasedness of the estimate conditional on the given history, and controls variance through the tail properties of the geometric distribution.
50%UnverifiedThrough the design of conditional geometric resampling and surrogate probabilities, the algorithm requires only a constant number of graph operations per update, thereby achieving nearly linear expected time without explicitly enumerating all unobserved nodes.
50%