March 6, 2020
Facebook debuted a powerful tool against spammers that uses Deep Entity Classification (DEC), a machine-learning technology. The company’s data science manager Bochra Gharbaoui reported the tool has already taken down 6.6 billion fake accounts last year and blocked millions of attempts to create new ones. DEC analyzes “deep features” of each individual profile, which refers to its behavioral patterns rather than its direct characteristics and includes the profile’s properties and groups/pages the user has contacted.
ZDNet reports that, “DEC doesn’t only register which groups a particular individual has joined; but also the number of admins in the group, its members, or the time since the group was created … [and not just] the number of friend requests sent by one profile, but also the number of requests sent by the accounts that this profile has befriended.”
The DEC algorithm aggregates these deep features “to find out things like the mean number of users in all the groups that one profile has joined, or the maximum number of groups that a given profile’s friends are part of.” With the 20,000 or so features gathered for each profile, said Facebook engineering manager Daniel Bernhardt, the tool can “see, through a number of signals, if a user is trying to misrepresent their identity.”
“It’s not so much about the content of an account, but about how that account interacts with others on the platform,” he explained.
Bernhardt added that the DEC’s main advantage is “how difficult it is to reverse engineer.” Other machine-learning methods that only look at a few features, he said, “are too easily tricked by attackers,” who figure out how to game the algorithm. Currently, anti-spam tools are trained on “a limited set of characteristics directly linked to the account.”
“The problem is adversarial robustness,” said Gharbaoui. “Adversaries can control direct features quite easily, for example by managing the number of friend requests they send out. That’s why you need to look at deep features — the graph and the model around the account.” She added that, “the estimated volume of fake accounts on Facebook is now 5 percent.”
ZDNet notes, however, that, “Facebook is nowhere near certain to win the war against fake accounts … [because] it is only a matter of time before attackers figure out a way around the improved tool.” “Adversaries move fast,” agreed Gharbaoui. “Their adaptation cycle is fierce, and it’s getting more sophisticated.”
For example, last year, Facebook “removed a global network of more than 900 accounts, pages and groups that had used sophisticated methods such as AI-generated profile pictures to spread pro-Trump narratives to 55 million users” — only a few months after the company removed 2,600 fake pages, groups and accounts involved in “coordinated inauthentic behavior.” For that reason, Bernhardt and Gharbaoui admitted that the field is still “very strongly attacker-controlled,” which means DEC will constantly “need re-training.”