HeyDonto launches DFT Labs to pursue physics-based machine learning capabilities

HeyDonto launches DFT Labs to pursue physics-based machine learning capabilities Artificial intelligence startup HeyDonto AI Technology today announced that it has established DFT Labs, a research subsidiary dedicated to pursuing a physics-based framework for machine learning. The launch of the new company follows the publication of the framework’s peer-reviewed foundational paper “Data Field Theory: A Geometric Framework for Learning on Riemannian Manifolds with Synthetic Validation and Limitation Analysis” in the academic journal Frontiers in Big Data, authored by Reza Nehzati, founder and chief scientific officer of DFT Labs. The paper is available online. The company argues that intelligence can be formulated as a mathematical framework in which learning can be represented by a continuous field on a Riemannian manifold. The paper and its underlying concepts borrow tools from physics that describe fields as curved geometric surfaces, commonly used to study magnets, superconductors and other systems. “We’re focused on understanding the relationship between all data points, and we believe that’s where a kind of intelligence forms,” Chief Executive Rivers Morrell told SiliconANGLE in an interview. He described the end result as intelligence that emerges from relationships rather than from plotting data. “We don’t study 100 birds and figure out how birds do formation. We study the formation and the relationship between the birds.” The analogy describes the advantage DFT Labs wants to demonstrate; the question that follows is whether relationships can be pulled out of real-world data that prove useful for machine intelligence and learning. The paper reports that the underlying mathematics achieved 89.2% accuracy on data classified using a manifold model, which outperformed the comparison methods used in the experiment. However, when researchers used MNIST digits, small grayscale scans of handwritten numbers used for image recognition testing, accuracy fell. After converting the digits into data points, DFT correctly classified 15.7% of them, only slightly above the 10% expected from random guessing and nearest-neighbor achieved 51.7%. These results suggest that DFT performs well on synthetic data designed around its geometric assumptions, but struggles when applied to real-world data whose underlying geometry is not known in advance. The paper identifies a key next step as enabling DFT to discover or learn the appropriate geometry directly from a dataset. HeyDonto said this paper is only the beginning; it opens as a foundational treatise. HeyDonto intends to show that DFT can measurably improve the AI models it is layered onto using real-world problems, demonstrating that this challenge can be overcome. Putting DFT Labs’ research to use The company has already put what it has learned from DFT Labs into production to underlie its current data curation and intelligence product, Axiomera. This is the company’s shared semantic-intelligence platform, used to turn fragmented enterprise data into standardized, harmonized AI-ready information. This same platform powers Conduit, the company’s dental and medical interoperability app, and Quantara, a clinical intelligence application for cancer research. As HeyDonto’s healthcare division, Quantara’s work includes structuring pathology, imaging and molecular information for clinical use. The company’s executives also said it has potential for administrative and pharmaceutical applications, such as inventory management, rebate analysis and data harmonization between research institutions. “Creating better, more accurate and faster algorithms for biomarker identification [can] help cancer patients get the right treatment faster,” Quintara President Kristopher Fuhr told SiliconANGLE. Fuhr added that HeyDonto’s commercial use cases generally follow peer-reviewed research findings. This positions DFT Labs as more than just an exercise in branding and hype, but as foundational technology with practical applications. Morrell added that the company intends to build DFT Labs’ research into robust applications of intelligence that can transform how large language model architectures work. HeyDonto hopes to build the framework into a foundational model that can compete against major model developers. “We are going to take on Anthropic and the big guys with our model,” said Morrell. With the launch of DFT Labs, the company plans to use academic benchmarks as opening proof that the data model and the potential architecture can stand up to real-world rigor. Morrell said that in about 12 months, the company hopes to release the benchmarks from industry and academic competition entries to give a confident display of applied use of the research. Image: SiliconANGLE/Microsoft Designer A message from John Furrier, co-founder of SiliconANGLE: Support our mission to keep content open and free by engaging with theCUBE community. Join theCUBE’s Alumni Trust Network, where technology leaders connect, share intelligence and create opportunities. 15M+ viewers of theCUBE videos, powering conversations across AI, cloud, cybersecurity and more 11.4k+ theCUBE alumni — Connect with more than 11,400 tech and business leaders shaping the future through a unique trusted-based network. Are you AWS customer? Support SiliconANGLE Financially by buying your AWS services from our Marketplace portal page and links. About SiliconANGLE Media SiliconANGLE Media is a recognized leader in digital media innovation, uniting breakthrough technology, strategic insights and real-time audience engagement. As the parent company of SiliconANGLE, theCUBE Network, theCUBE Research, CUBE365, theCUBE AI and theCUBE SuperStudios — with flagship locations in Silicon Valley and the New York Stock Exchange — SiliconANGLE Media operates at the intersection of media, technology and AI. Founded by tech visionaries John Furrier and Dave Vellante, SiliconANGLE Media has built a dynamic ecosystem of industry-leading digital media brands that reach 15+ million elite tech professionals. Our new proprietary theCUBE AI Video Cloud is breaking ground in audience interaction, leveraging theCUBEai.com neural network to help technology companies make data-driven decisions and stay at the forefront of industry conversations.