Check formulation of attribute inference metric

[jim-smith]

we currently report the ratio of the proportion of training records that. are vulnerable to the proportion of test records that are vulnerable

if we're to. take the logarithm of that it would be the difference of the logs of the likelihood that something can be imputed. AS far as I recall this is then becoming similar to the way that the LIRA characterises the 'likelihood ratio' for it's MIA attack. Can we check and if it is similar consider reporting it in a similar formulation