Conceptlikelihood ratio

Likelihood Ratio

The basic object of differential privacy accounting. The logarithm of the ratio of output probabilities for adjacent datasets is the privacy loss, and its distribution determines the (epsilon, delta)-DP guarantee.

Timeline (last 90 days)

Oct 19

Under the e-value framework, the likelihood ratio is itself an e-value and is optimal in certain senses (for example, it has the largest expected log growth under the alternative hypothesis).

Unverified50%

All Facts (1)

Unverified

Under the e-value framework, the likelihood ratio is itself an e-value and is optimal in certain senses (for example, it has the largest expected log growth under the alternative hypothesis).

50%

Source Articles

Multi-scale FFT Accelerates Privacy Loss Accounting for Random Allocation SamplingNov 2