Early Detection of Online Fraud Using Unsupervised Machine Learning—Fang Yu—DataVisor

“Existing solutions like supervised machine learning are very reactive to what the attacker is currently doing, so there’s always a cat and mouse game; the unsupervised machine learning that we are building….

doesn’t have existing assumptions of what the fraud looks like, and that itself is actually more robust in terms of detection,” says Fang Yu, CTO and co-founder of DataVisor, a company whose mission is to protect large social and financial institutions from the increasing number of sophisticated cyber-attacks.

When the nature of cyber-attacks can literally change on an hourly basis, the technology which relies upon the characteristics of past fraudulent behavior is insufficient.

DataVisor’s technology mitigates this problem by reviewing billions of accounts and identifying patterns of fraudulent behavior.

Yu offers a detailed and informative conversation about DataVisor’s services, which are currently protecting over four billion user accounts globally.

She also discusses the advanced techniques used by fraudsters, the importance of early fraud detection, and an upcoming enterprise version of their technology which will allow clients to adjust the models themselves and have more leverage over the algorithms.

Tune in, learn more by visiting datavisor.com, and email your questions to info@datavisors.com.

Podcast: Play in new window | Download | Embed