GLYNT.AI, which extracts unstructured data from complex documents, announced the integration of Low Shot Learning into its machine learning system. Low Shot Learning reduces the size of the training data set for document content extraction and classification to less than 7 documents, as compared to the 15,000 to 300,000 documents needed for training by other machine learning systems. With Low Shot Learning, customers can transform unstructured data into clean, labeled data in under an hour.
Low Shot Learning is a product feature that results from a specific configuration of the machine learning algorithms. Coupled with high accuracy rates, typically 96 – 98%, and an Elastic AI Workbench, the GLYNT machine learning system now easily provides high-quality results with negligible setup costs and complete scalability.
“Low Shot Learning is a game changer along several business objectives,” said Martha Amram, CEO of GLYNT.AI. “First, it takes out time and money from the extraction of unstructured data. What previously took a data science and engineering team 8 – 15 months can now be done in under an hour by a single office worker.
Second, there is little performance risk. Even with the acceleration from Low Shot Learning, accuracy rates of 96 – 98% are maintained. Also, our self-contained engineering infrastructure, the Elastic AI Workbench, provides complete scalability. GLYNT is available as a cloud-based service or can be deployed behind our customers’ firewall within a day.”
“We’re also excited to add a powerful upside to the business case for data liberation,” Amram continued. “Typically, the data wrangling and training tasks are 90% of the cost of a machine learning endeavor. When Low Shot Learning turns months of work into just an hour, unstructured data extraction has a solid ROI for smaller data sets, too, vastly increasing the market.”
Sandra Carrico, VP of Engineering and Chief Data Scientist remarked, “Our high-performing and highly-scalable system, now with Low Shot Learning, is the result of deep experience in extracting unstructured data from complex documents. We use the GLYNT system in two of our energy industry products and have now added other verticals. GLYNT is a general purpose solution, and works well in many settings. Its performance is not sensitive to the nature of the content, and also handles multiple file types (pdfs, scans or images) within the same system. We know the pain of current solutions, so we built the system that solved these real-world challenges.”
Please contact info(at)glynt(dot)ai for more information.
GLYNT is a machine learning system in commercial use that produces a stream of clean, labeled data from any document. Get started in minutes. GLYNT was developed by the team at WattzOn, which uses GLYNT in its products for the energy and credit markets. Four key product features make GLYNT a powerful and easy-to-use product: Low Shot Learning provides results in under an hour; the Elastic AI Workbench enables complete scalability; the customer-facing web tool provides transparency and control; and the high level of accuracy, 96 – 98%, enables GLYNT to be used in demanding applications such as invoice processing, healthcare records and loan documents.