Facebook AI Framework Doubles Fake Accounts Detection in 2019

Share Article

Artificial Intelligence has had its fair share of criticism, but very can deny its use case in businesses, particularly with companies that deal with large volumes of data. In its recently published Community Standards Enforcement Report, Facebook explains that it has been able to root out up to 3.2 billion fake accounts on its platform between April and September 2019 using its AI framework.



DEC to the Rescue

According to Facebook, the number of fake accounts removed was well over the 1.5 billion fake accounts flagged in the same period last year. The company attributes the growth in discovery efficiency to the integration of the Deep Entry Classification (DEC), a machine learning framework developed by the company.

The DEC is a framework for detecting abusive accounts. The company shed light on the AI for the first time at its Fighting Abuse Scale 2019 event in Menlo Park earlier this month. According to the social media giant, the framework improves its machine-learning account segmentation technique. In the past, it was difficult for Facebook to verify real accounts in isolation, and the models created by engineers were not sophisticated enough, making it easier for criminals to game the system.

Thus, the DEC uses a “multistage, multitask-learning paradigm that leverages a large number of medium-precision, automated labels and a small number of high-precision, human-labeled samples” to detect fake accounts from real ones.

Facebook Cautiously Approaching AI

As expected, fake account detection is far from the only thing that Facebook has been using AI to optimize. The Silicon Valley giant is one of the tech companies that have dedicated themselves to building on this innovative technology, deploying innovative solutions to grow its operation.

Currently, Facebook has deployed AI algorithms on its core platform, as well as throughout its family of apps- WhatsApp, Instagram, and Messenger. Still, the company appreciates that they have a lot of work to do to develop the technology up to a point where it will be able to “match human intelligence.” In an interview with WIRED Magazine, Facebook’s Vice President of Artificial Intelligence, Jerome Pesenti explained that while they’ve had some progress in integrating AI, there is still a significant distance to walk.

“Deep learning and current AI, if you are really honest, has a lot of limitations. We are very very far from human intelligence, and there are some criticisms that are valid: It can propagate human biases, it’s not easy to explain, it doesn’t have common sense, it’s more on the level of pattern matching than robust semantic understanding,” he said, while also noting that they were confident in their objective to see this technology develop into something that can be applied to every facet of human life.


Images are courtesy of Shutterstock, Twitter.

Share Article

Related topics

Follow Author

Daily signals, Bitcoin analytics and traders chat. Join our Telegram today!

Let’s Go
Daily signals and Bitcoin analytics.