Deep learning, a technique falling under the umbrella of artificial intelligence (AI), has rapidly gained popularity recently as a way to automate fraud detection through real-time insights. With deep learning, complex algorithms are trained to recognize the signs of potentially fraudulent activity, enabling users to not only identify anomalous or suspicious behavior but also prevent imminent threats. When powered by high performance computing (HPC) solutions, the immediate nature of this valuable information means financial companies can not only identify fraud that is in progress, but identify and take steps to prevent imminent threats.
In this highly digital age, today’s consumers manage nearly their entire lives online. In the course of a day, many consumers will come in contact with a number of web-based financial platforms, including online banking portals, wealth management systems, online payment systems, and e-retailers. As consumers increasingly take advantage of web-based or mobile platforms to execute and manage financial transactions, it has become a fact of life that sooner or later everyone is likely to be affected by credit card fraud.
A recent fraud study by ACI Worldwide found that 46% of Americans have had their credit card information compromised within the last 5 years, while another study placed the current domestic losses resulting from credit card fraud at approximately $16 billion in 2016 alone. For example, Equifax’s recent data breach affected 145.5 million Americans, exposing sensitive personal information. Savvy financial companies must continuously find new ways to gain faster intelligence into fraudulent activity, so they can stay ahead of criminal activity and maintain controls for economic crime.
The chief challenge is that deep learning is a highly compute-intensive exercise, and legacy technologies are unable to deliver the extensive compute power and scalability needed to fully realize the business benefits of deep learning. With the proliferation of AI capabilities, NVIDIA® graphics processing units (GPUs) and HPC solutions have become crucial to exploit these market-changing technologies, and enable businesses to discover the effect of real-time data analytics for data security.
HPC leader Hewlett Packard Enterprise (HPE) is enabling financial companies to build more robust compute platforms focused on redefining data security and helping businesses gain intelligence to better protect themselves. To achieve this, they have teamed up with NVIDIA, the AI company, and Kinetica, a software application provider, to develop a solution that automates real-time fraud detection with GPU computing. GPU-accelerated computing works to optimize AI workloads to deliver faster, more accurate intelligence to keep businesses safe. The new performance-optimized, cost-effective solution is designed specifically to help businesses secure credit card transactions with real-time fraud detection capabilities.
Many of today’s fraud systems are built using a multitude of different tools, and data must be moved between different systems for various types of analysis. This solution leverages Kinetica’s GPU-based analytics software for in-database processing performance and efficiency, which helps financial businesses safeguard sensitive customer information and dramatically reduces the risk of data loss.
The rise of online platforms and new, powerful computing technologies means there are a growing number of ways that criminals can target financial institutions and their customers. Financial businesses need to take steps to protect themselves and leverage game-changing technologies like deep learning to automate the very complex task of credit card fraud detection.
Together, HPE and NVIDIA are working to empower customers with real-time insights by harnessing the extensive acceleration power through NVIDIA GPUs. To learn more about this solution and deep learning for fraud detection, I invite you to follow me on Twitter @Bill_Mannel. And to stay on top of the latest news and solutions for HPC, AI/deep learning, supercomputing and more, you can also check out @HPE_HPC and @NvidiaAI.