<a​ck-j:matrix.org> De-anonymizing Monero: A Maximum Weighted Matching-Based Approach.

<a​ck-j:matrix.org> moneroresearch.info/index.php?actio…n=resource_RESOURCEVIEW_CORE&id=260

<a​ck-j:matrix.org> This paper has a flashy title but they use the ML dataset I published and achieved roughly the same accuracy with a ring size of 11 (slightly above random guessing). Nice to see that my approach was validated by a different group and no significant privacy leak was discovered in the metadata.

<r​ucknium:monero.social> xmrack: On a very quick skim, it looked like they tried to use a graph attack, too. Is that correct? Like the DM decomposition paper.

<r​ucknium:monero.social> AFAIK, your paper only used, at most, one hop backward in the tx graph.

<a​ck-j:matrix.org> I used two hops IIRC but its been a while

<r​ucknium:monero.social> I'll try to read their paper in full before next meeting.

<a​ck-j:matrix.org> matrix.monero.social/_matrix/media/…matrix.org/dUGnAyljDSxMpmeBjCiNAYco

<a​ck-j:matrix.org> I used two hops but labeled them as hop 0 and hop 1

<r​ucknium:monero.social> This is the second paper that has used your dataset, right? Beside your own paper, of course.

<a​ck-j:matrix.org> Correct, i’m pretty happy about that :)