Data privacy regulations such as GDPR, CCPA, and HIPAA present a challenge to training AI systems on sensitive data, such as financial transactionsand patient health records and user device records. Historical data is what “teach” AI systems to identify patterns and make predictions, but there are technical hurdles to using it without compromising a person’s identity.
One solution that has gained popularity in recent years is standardized learning. Technology It trains a system across multiple devices or servers that hold data without ever exchanging it, enabling collaborators to build a shared system without sharing data. Intel recently partnered with Penn Medicine to develop a brain tumor grading system using standardized learning, while a group of major pharmaceutical companies, including Novartis and Merck, have built a unified educational platform to accelerate drug discovery.
Tech giants, including Nvidia (via Clara), offer Unified Learning as a Service. But a new startup, DynamoFL, hopes to challenge incumbents with a federated learning platform focused on performance, ostensibly without sacrificing privacy.
“DynamoFL was founded by two MIT PhDs in electrical engineering and computer science, Christian Lau and myself, who has spent the past five years working on privacy-preserving machine learning and hardware for machine learning,” CEO Vaikunth Muguntan told TechCrunch in an email interview. We discovered a huge market for federal learning after we received frequent job offers from leading finance and technology companies that were trying to build federated learning internally in light of emerging privacy regulations such as the General Data Protection Act (GDPR) and Consumer Privacy Protection Act (CCPA). During the process, it was clear that these organizations were struggling to stand up to federal learning internally and we built DynamoFL to address this market gap.”
DynamoFL – which they claim to own Major clients in the Automotive, Internet of Things and Finance sectors – still in the early stages of their go-to-market strategy. (The startup currently has four employees, with plans to hire 10 by the end of the year.) But DynamoFL focused on optimizing new AI technologies to stand out from the competition, offering capabilities that supposedly boost system performance during Fighting Attacks and weaknesses in federated learning – such as “member heuristics” attacks that make it possible to discover data used to train a system.
“Our Personal Federal Learning Technology… Enabling[s] Machine learning teams fine-tune their models to improve the performance of individual groups. This gives C suite executives higher confidence when deploying machine learning models that were previously considered black box solutions” Mugontan said. “this is [also] It sets us apart from competitors like Devron, Rhino Health, Owkin, NimbleEdge, and FedML who suffer from the common challenges of traditional federated learning.”
DynamoFL also advertises its platform as being cost-effective versus other privacy-preserving AI point solutions. sFederal learning doesn’t entail massive data aggregation on a central server, DynamoFL can lower data transmission and computation costs, Moguntan asserts — for example, allowing a customer to send only small, incremental files instead of petabytes of raw data. As an added advantage, this can reduce the risk of data leakage by eliminating the need to store large amounts of data on a single server.
“Common privacy-enhancing techniques such as differential privacy and standardized learning have suffered from a perpetual ‘privacy versus performance’ trade-off, in which the use of more robust techniques to maintain specificity during model training leads to poor model accuracy. “This serious challenge has prevented many machine learning teams from adopting the privacy-preserving machine learning technologies required to protect user privacy while complying with regulatory frameworks,” said Mouguntan. “DynamoFL’s custom Unified Learning solution addresses a critical obstacle to machine learning adoption.”
Most recently, DynamoFL closed a small seed round ($4.15 million at a $35 million valuation) involving Y Combinator, Global Founders Capital and Basis Set; The startup is part of Y Combinator’s Winter 2022 group. Mouguntan says the proceeds will primarily go toward hiring product managers who can integrate DynamoFL’s technologies into easy-to-use future products.
The pandemic has highlighted the importance of rapidly leveraging diverse data for emerging crises in healthcare. In particular, the pandemic has emphasized the need for greater access to critical medical data in times of crisis, while maintaining patient privacy,” Mouguntan continued. “We are well positioned to overcome the slowdown in technology. We currently have three to four years of runway, and tThe ech slowdown has helped our recruitment efforts. The biggest tech companies were hiring the majority of the leading federal learning scientists, so the slowdown in big tech hiring has provided an opportunity for us to hire the best talent in federal and machine learning.”