SOURCE: Saffron


December 18, 2014 13:17 ET

Saffron Granted U.S. Patent for Hyper-Sparse Matrix Compression

Innovative Technique Enables More Efficient Storage and Distribution of Knowledge, in Cognitive Computing Platform

LOS ALTOS, CA--(Marketwired - December 18, 2014) - Saffron, the first artificially intelligent, cognitive computing platform that learns, reasons and anticipates like you and for you, today announced it has been awarded U.S. Patent #8,909,609, "Methods, systems and computer program products for providing a distributed associative memory base." Saffron focuses on helping companies within the industrial internet optimize product lifecycle management and personalized customer experience by knowing the customer individually, providing faster insight to defect and issue resolution, and insuring that the knowledge of one is available to many.

The patent is the second of two awarded for Saffron's Memory Base, which defines the patterns from data using statistical dependencies between the dots over time to anticipate what is not yet in data, similar to our brain using past experiences to anticipate future outcomes. This newer patent is focused specifically on a technique to compress the memory base so that it can be accessed and modified more efficiently, saving time and reducing total cost of ownership for Saffron's customers.

The inventors were Saffron's Chief Software Engineer, President and Co-Founder Jim Fleming along with Yen-min Huang, the company's Chief Scientist. The patent was initially filed in June of 2009, and the technology described in the patent has been in use for several generations of Saffron's platform.

"Saffron can ingest and process enormous amounts of data, but equally important is distributing the knowledge created from this data, so customers can quickly and contextually access this knowledge while in the field," explained Saffron CEO Gayle Sheppard. "Since we utilize lossless compression techniques and support real-time, instant learning, Saffron memories can be retrieved or edited without decompressing and recompressing, delivering a knowledge store efficiency for our customers that is an order of magnitude beyond what other solutions have attempted."

Saffron recently unveiled Saffron 10, the newest version of its groundbreaking Natural Intelligence Platform. Saffron 10 fuses the power of computing with brain-like intelligence, enabling organizations to more quickly focus on the knowledge that matters to them most (including things they may not think to look for), anticipate what will happen next, and optimize decisions based on that criteria.

Saffron deploys a patented associative memory or "natural learning" approach. It finds connections among data across diverse sources without the need for rules or modeling, while learning incrementally and anticipating outcomes based on patterns it finds in the data. Designed to be easily integrated with existing investments and initiatives, Saffron 10 ingests data natively from databases, file systems, and common file formats, making it seamless for businesses to apply the power of Saffron to other applications.

Saffron is commercially deployed for real-time operational customer and product lifecycle intelligence and decision support in smart device manufacturing, financial services, energy and national security industries. Saffron Operational Intelligence applications are numerous and include personalized 1:1 customer product and service prediction, model-free fraud, waste and abuse intelligence, personalized predictive maintenance, country risk assessment and threat analysis.

For more information on Saffron's Memory Base and its Matrix Compression technique, please review our technical white paper, Your Brian is Cognitive, Not a Database.

Patent Abstract

Systems, methods and computer program products are provided for a distributed associative memory base. Such methods may include providing a distributed memory base that includes a network of networks of associative memory networks. The memory base may include a network of associative memory networks, a respective associative memory network comprising associations among a respective observer entity and a plurality of observed entities that are observed by the respective observer entity. Ones of the associative memory networks are physically and/or logically independent from other ones of the associative memory networks. Methods include imagining associations from the associative memory base using a plurality of streaming queues that correspond to ones of a plurality of rows of ones of the associative memory networks.

Saffron is the first cognitive computing platform that learns, reasons and anticipates like you, for you. The platform learns instantly from multi-sourced and types of data to adapt in real time, automatically connecting the dots to illuminate the knowledge that really matters and to help you anticipate what will happen next. Saffron enhances the speed and volumes at which data can be processed but also critically improves the accuracy of results. Businesses using Saffron can anticipate market trends, optimize processes, mitigate risk, personalize customer experiences and find new revenue streams. Founded in 1999, Saffron is headquartered in Los Altos, California. For more information, please visit

Contact Information