DynamoFL aims to bring privacy-preserving AI to more industries

Published by
Peter Kavinsky

Data privacy regulations such as the GDPR, CCPA and HIPAA pose challenges for training AI systems to deal with sensitive data such as financial transactions., patient medical records, and user device logs. Historical data is what “teaches” AI systems to spot patterns and make predictions, but there are technical barriers to using them without compromising human identity.

One workaround that has become popular in recent years is federated learning. Technique trains the system on multiple devices or servers containing data without exchanging them, allowing employees to create a common system without exchanging data. Intel recently partner with Penn Medicine to develop a classification system for brain tumors using federated learning, while a group of major pharmaceutical companies, including Novartis and Merck, built federated learning platform to accelerate drug discovery.

Tech giants including Nvidia (via Clara) offer federated learning as a service. But a new startup DynamoFLhopes to compete with incumbents with a federated learning platform that focuses on performance, ostensibly without sacrificing privacy.

DynamoFL was founded by two MIT Electrical and Computer Science PhDs, Christian Lau and myself. who has spent the last five years working on privacy-preserving machine learning and machine learning hardware,” CEO Vaikkunt Muguntan said in an email interview with TechCrunch. “We discovered a huge market for federated learning after we received repeated job offers from leading financial and technology companies that were trying to create federated learning in-house in light of new privacy regulations such as the GDPR and CCPA. During this process, it became clear that these organizations were struggling to support federated learning in-house, and we created DynamoFL to fill this gap in the market.”

DynamoFL which claims that key customers in the automotive, IoT, and financial sectors are in the early stages of their go-to-market strategy. (The startup now has four employees and plans to hire 10 by the end of the year.) But DynamoFL has focused on refining new AI techniques to stand out from the competition, offering features that purportedly improve system performance while at the same time struggle attacks and vulnerabilities in federated learning – such as “member inference” attacks that allow the discovery of data used to train a system.

Image credits: DynamoFL

“Our personalized federated learning technology… allows[s] machine learning teams to fine-tune their models to improve the performance of individual cohorts. This gives senior executives more confidence in deploying machine learning models that were previously considered black-box solutions.” Muguntan said. “This [also] distinguishes us from competitors such as Devron, Rhino Health, OakinNimbleEdge and FedML that struggle with the common problems of traditional federated learning.”

DynamoFL also touts its platform as being cost effective compared to other privacy-preserving spot AI solutions. FROMBecause federated learning doesn’t require bulk data collection at a central server, DynamoFL can cut data transfer and compute costs, says Muguntan, for example by allowing a client to send only small incremental files rather than petabytes of raw data. As an added benefit, this can reduce the risk of data leakage by eliminating the need to store large amounts of data on a single server.

Common privacy enhancement technologies such as differential privacy and federated learning have suffered from a constant trade-off between “privacy and performance” where the use of stronger privacy practices during model training inevitably results in reduced model accuracy. This critical bottleneck issue has prevented many machine learning teams from implementing the privacy-preserving machine learning technologies needed to protect user privacy while complying with regulatory requirements,” Muguntan said. “DynamoFL’s personalized federated learning solution breaks a critical barrier to machine learning adoption.”

DynamoFL recently completed a small seed round ($4.15 million at a $35 million valuation) that included Y Combinator, Global Founders Capital and Basis Set; the startup is part of the Y Combinator Winter 2022 package. Muguntan says the proceeds will primarily be used to hire product managers who can integrate DynamoFL technologies into future user-friendly products.

“The pandemic has highlighted the importance of rapidly leveraging diverse data to address emerging health crises. In particular, the pandemic has shown how critical medical data needs to be made more accessible during a crisis while maintaining patient privacy,” continued Muguntan. “We are well positioned to weather the slowdown in technology. We currently have three to four years of runway, and tThis slowdown has actually helped our hiring efforts. The biggest tech companies were hiring most of the top federated learning scientists, so the slowdown in hiring in big tech has given us the opportunity to hire the best in federated learning and machine learning.”

Peter Kavinsky

Peter Kavinsky is the Executive Editor at cablefreetv.org

Recent Posts

  • News

The last party of the queen of Colombian folklore Totó La Momposina

JFIF Firmware, Firmware ¨http://ns.adobe.com/xap/1.0/ 11003000:Politics:Elections11006000:Policy:GovernmentcoThe last party of the queen of Colombian folklore Totó La…

5 mins ago
  • News

Scotland close in on Nations League promotion with upset win over Ireland

Scotland are one point away from promotion to League A in the Nations League after…

6 mins ago
  • News

Stranger Things final date

Research on the history of the new season of the Stranger Things series, which has…

8 mins ago
  • News

Scotland close in on Nations League promotion with win over Ireland

Click Here to Watch this Event Live Online for Free! Scotland are one point clear…

9 mins ago
  • News

Majima’s Last Stand, Mika and Yoshi’s Reunion, and the Aftermath of the Radio Tower Incident

Click Here to Watch this Event Live Online for Free! A few hours earlier, Recoil…

10 mins ago
  • News

OCC Theater Department Focuses on Coastal Cleanup – Coast Report & More News

OCC Theater Department Focuses on Coastal Cleanup coast report Today Latest News Update I have…

15 mins ago