Data Innovation & AI at Capital One with Adam Wenchel - #147

Fraud Detection

At Capital One, innovative AI-driven solutions are transforming fraud detection and prevention. Adam Wenchel highlights "SecondLook," a machine learning application that identifies suspicious transactions, such as unusually high restaurant tips or duplicate charges, to alert customers 1. This system relies on a balance of precision and user experience, ensuring that customers are notified of potential issues without being overwhelmed by false alarms.

It's really one of the kind of innovative ways that we're improving our customer experience.

--- Adam Wenchel

Additionally, Capital One employs unsupervised learning and anomaly detection to refine their fraud detection capabilities, leveraging customer feedback and data scientist expertise to enhance accuracy 2.

   

Cybersecurity

Capital One's approach to cybersecurity integrates machine learning to tackle complex threats. Adam Wenchel explains that while they utilize best-of-breed cybersecurity products, they also develop custom solutions to address unique challenges, such as detecting malware callouts using domain generating algorithms (DGAs) 3. This innovative work involves convolutional neural networks (CNNs) to identify suspicious domain names, distinguishing between legitimate and malicious ones.

It's continually, whether it's spamming or cybersecurity, it's a constant cat and mouse game.

--- Adam Wenchel

By applying CNNs, typically used in image recognition, to cybersecurity, Capital One enhances its ability to detect and prevent cyber threats, maintaining a proactive stance against evolving adversaries 4.