RT @SherriDouville: @ctlin1 Is someone going to address the reidentification that can be done with ML most of the time with deidentified da…
3,050 followers
1,098 followers
@ctlin1 Is someone going to address the reidentification that can be done with ML most of the time with deidentified data @ctlin1 👇 https://t.co/iq0LMTxZfW
62 followers
@tomstafford @sTeamTraen This paper may(!) give you some further bg (didn't read): https://t.co/yHuaFwfxSq and other techniques to score re-id risk. 3/
92 followers
Here is one for the #Privacynerds In general we found that the Pitman model emerged as the most accurate (for re-identification) for low sampling fractions. For higher sampling fractions the most accurate estimate varies between SNB and Zayatz.https://t.co