AI + HPC Aimed at GDPR, Data Privacy Compliance
In the banking, financial services and insurance industries, data privacy compliance is an eternally daunting challenge, and now data management consultancy Vaco and NEC X, the "innovation accelerator" arm of NEC, have integrated AI with high performance computing to handle PII (personal identification information) operations redaction within GDPR and other data privacy workloads faced by BFSI organizations.
NEC X, which fast tracks NEC Corp.’s emerging technologies, brings to the joint reference design its SX-Aurora Tsubasa AI platform, which uses vector processing to provide HPC-class performance (2.45 teraflops/second throughput) on a compact PCIe x16 card, according to Robbert Emery, director of technology marketing/business development.
SX-Aurora Tsubasa fits hardware environments ranging from workstation towers for a single card to rack-mounted servers for multiple cards, according to NEC X. Its Vector Engine (VE) processor was the first implementation of one processor with six HBM2 memory modules using chip-on-wafer-on-substrate technology, delivering memory bandwidth of 1.2 TB/s. Emery told us the system utilizes the MPI interface between cards and Infiniband between multiple servers.
Vaco brings its Cognitive Computing Framework (CCF) with algorithms built for GDPR, CCPA, ISO/IEC 38500 or other regional data protection regulations. The companies said the offering can be scaled by number of VE cards as needed to process large data sets in real time for compliance workloads.
Sudhir Sahu, Vaco partner who leads the consultancy’s banking and financial services practice, told us CCF software has been built on top of SX-Aurora Tsubasa “specific to PII… and we have built a GDPR compliance rules engine using machine learning and AI. Before we push the data into a data lake to store it, we redact the PII data that might be sitting in an email, or in a Word document, Excel spreadsheet or PDF attachment. Some people send their social security number, their credit card number, driver’s license number, date of birth, etc. And those, as per the federal government's requirements for the banking industry, have to be redacted before being stored into a data lake to meet compliance requirements.”
Compounding the challenge is scale – data volumes beyond the processing capacity of conventional servers, Sahu said. Banks send and receive millions of emails per day, “so without the combination of this hardware power that NEC X has built, and the machine learning capability that we have to create this application, we would not be able to process the amount of data that has to be processed within a few seconds.”
Developed over the past six months, the platform has been implemented by two early, undisclosed, adopter banks, Sahu said. The integration is intended to provide a solution that’s “80 percent ready for deployment out of the box.” The remaining 20 percent is custom programming required to comply with data governance regulations within customers’ specific countries. The offering is intended to improve the precision of financial rules and models, accelerate documentation generation to meet privacy and regulatory needs, enable predictive analytics across industries, interpret customer behaviors and the capability to ingest and analyze large volumes of structured and unstructured data.
“This partnership combines our affordable and scalable AI platform with Vaco’s domain expertise in IT management services and big data software development for regulatory compliance, providing enterprises with a more secure way to comply with regional data regulations,” said Shige Ihara, CEO of NEC X. “Together, we can dramatically reduce the implementation time to as little as one month, while saving IT departments 10x in CAPEX costs and providing a 20 percent smaller footprint with a 20x computational improvement, compared to traditional high performance computing (HPC) servers.”