Industry Leaders Gather with Consilient and Intel Executives to Discuss How Federated Learning Unlocks the Full Potential of Artificial Intelligence for Fighting Financial Crime at Federated Learning Conference
Consilient and Intel team up to teach about federated learning’s tremendous impact on the future of business and society at the Second Annual Federated & Distributed/Decentralized Machine Learning Conference ARLINGTON, Va. This week, Consilient and Intel executives led three educational discussions on federated learning and its impact for the anti-money laundering (AML) and financial crime detection industry with academic and industry leaders at the Second Annual Federated & Distributed/Decentralized Machine Learning Conference.
Keynote presentations addressed the basics of federated learning, its rapid growth and complexities, and the technical challenges that need to be met in order for federated learning to truly meet the goal of successful decentralized artificial intelligence (AI).
Juan C. Zarate, co-founder and chair of Consilient, and Mike Blalock, general manager of the financial services industry at Intel Corporation, led a keynote presentation entitled “A Revolutionary Moment for Federated Machine Learning in Fighting Financial Crime.” Zarate and Blalock gave a high-level framework of the current federated learning space and discussed how federated learning allows entities to effectively and accurately use data across organizations while meeting regulatory, privacy, and security requirements.
“We are in a critical, revolutionary moment in the financial regulatory and marketplace. This is not hyperbole – this is an intentional marker of where we think we are with this technology that improves financial crime risk management,” remarked Zarate. “Today’s hugely inefficient system allows nefarious actors to profit from their illicit behaviors. That is why we are redesigning the system with federated learning, which allows us to respect data localization laws and privacy concerns while expanding data access.”
“While federated learning shares algorithms across financial institutions, there needs to be sufficient protection. Confidential computing is a game changer because when used in the federated learning space, it achieves a delicate balance of data privacy with data innovation by keeping data and IP confidential,” said Blalock.
Gary M. Shiffman, Ph.D., co-founder and CEO of Consilient and founder and CEO of Giant Oak, and Nikhil Deshpande, Ph.D., director of AI and Security Solutions at Intel, led a deep dive session on applied federated machine learning entitled “Fighting Financial Fraud with Privacy-Preserving Federated Learning.” Shiffman and Deshpande addressed the shortcomings of the existing financial crime detection system and offered a technological explanation of how Consilient’s Dozer technology coupled with Intel’s SGX Trusted Execution Environment addresses security and privacy challenges.
“The global spend on legacy AML systems is massive and ineffective. There is a $180 billion annual spend on the global fight against financial crime, but less than 1% of financial crime is actually detected. Most AML systems deployed around the world are producing 95% false positives,” explained Shiffman. “The solution we built is the first-ever federated machine learning application for anti-money laundering. When deployed into financial institutions, our flagship product, Dozer, enables collaboration between financial institutions in a way that lowers false positive rates, increases discovery of threats, and therefore enhances the resiliency of the entire banking system while preserving privacy.”
Deshpande made the point that certain protections must be in place to ensure security while using federated machine learning. “We need to address threats of model tampering, theft and data loss in the system of federated learning for broader adoption. Intel® SGX delivers the smallest potential attack surface of any Trusted Execution Environment (TEE) available for the data center and helps increase protection against model tampering and data loss in federated learning,” said Deshpande.
Parviz Peiravi, CTO of Financial Services Industry Solutions at Intel, moderated the conference’s keynote panel entitled “Application of Federated Learning in Finance and Regulated Industries.” Panelists Angelena Bradfield, senior vice president of AML/BSA and Sanctions & Privacy at the Bank Policy Institute, Rick Hamilton, senior vice president of AML Risk Management at PNC Bank, and Zarate discussed the business, regulatory and technology applications of federated learning and explained why regulatory industries are the perfect match for federated learning.
“Both regulators and policymakers have bought into the need for AML reform and that technological innovation is essential to realizing an effective U.S. AML regime. Financial institutions should seize the moment and, in partnership with regulators, explore technologies that could enhance their AML compliance programs,” said Bradfield, addressing the need for regulatory and technology experts to work together.
“Tech companies have historically seen regulators, whose priority is to comply with policies, as barriers to adopting essential technologies. But in the last five years, we made significant progress; we have been working with regulators and the financial services industry together to solve major challenges. As an example, federated learning technology brings the current AML system into a broader view and moves the industry beyond analyzing the data set of a single bank. Instead, we can analyze the data sets of multiple banks without moving data while preserving privacy and thereby equip financial institutions with better tools to identify fraudulent activities and comply with regulations,” said Peiravi.
“Financial institutions need robust systems that effectively identify unusual transactions. Due to the sheer volume of transactions processed by financial institutions, technology investments are a fundamental part of any successful AML program,” explained Hamilton. “If the financial services industry can leverage federated learning to quantify and share criminal typologies, then we will be able to enhance the global AML effort by creating more effective and efficient behavior detection models while protecting customer and competitive information.”
This conference is one of many initiatives from Consilient and Intel to equip industry leaders to efficiently and effectively combat financial crime using federated learning. Shortly after the launch of Consilient in October 2020, Consilient and Intel published a white paper on federated learning entitled “Federated Learning through Revolutionary Technology,” which can be viewed here.
For more resources on the use of federated learning to combat financial crime, please visit www.consilient.com.
Founded in 2020 through a partnership between K2 Integrity and Giant Oak, Consilient brings together next-generation technology and best-in-class anti-money laundering and countering the financing of terrorism (AML/CFT) knowledge. Consilient is proud to be collaborating with Intel to bring this model to the market. This combination powers a more secure, dynamic and effective solution for financial institutions.