Saffron Technologies develops cognitive computing systems that use incremental learning to understand and unify by entity (person, place or thing) the connections between an entity and other “things” in data, along with the context of their connections and their raw frequency counts.
This approach provides a semantic and statistical representation of knowledge. Saffron learns from all sources of data including structured and unstructured data to support knowledge-based decision making. Its patented technology captures the connections between data points at the entity level and stores these connections in an associative memory
Associative memory bases identify entities that are defined as people, places and things. The technology mimics human memory by recalling associations between those people, places and things, and specifically the context and frequency of association. As each entity in a set of data has its own memory about all the other entities it is associated with, the engine can learn the way humans do.
The technology has proved to be highly scalable and efficient in cases of extremely large datasets. Saffron exploits a proprietary loss less compression routine that is capable of creating extremely compact models. The Saffron implementation is able to operate on compressed datasets thereby enabling dramatic reduction in storage and CPU hardware, thus enabling the application of associative memory technology in a distributed environment.
During the Iraqi insurgency Saffron was used by coalition forces in Iraq to help identify entities that were defined as people, places and things involved in insurgent networks. The tool mimicked human memory by recalling associations between those people, places and things. Unlike other tools Saffron focused on context and frequency of association.
Intel Saffron Anti-Money Laundering (AML) Advisor uses explainable AI to enhance decision-making for investigators and analysts. Associative memory AI finds and explains multidimensional patterns so that investigators and analysts can explore emerging trends across a bank’s or insurer’s data.
With an unsupervised learning approach, the AML Advisor unifies structured and unstructured data from enterprise systems, email, web and other data sources to deliver insights along with the explanation of how connections were identified.
Intel Saffron’s associative memory AI simulates a human’s natural ability to learn, remember and reason in real time. It mimics the associative memory of the human brain to surface similarities and anomalies hidden in dynamic, heterogeneous data sources, while accessing an infinitely larger data set than its human counterparts.
The AML Advisor surfaces these patterns in a transparent way, paving the way for “white box AI” in enterprise applications. These solutions are designed to enhance decision-making in highly complex tasks, and early results indicate they can catch money launderers with unprecedented speed and efficiency.
Total financial crime is at all-time highs. According to the United Nations, the estimated amount of money laundered globally in one year is 2 to 5 percent of global GDP, or $800 billion to $2 trillion. In addition, in 2016 alone, approximately 15.4 million consumers were victims of identity theft or fraud, resulting in $16 billion in losses.
Different than Deep Learning
It takes a lot of training for Deep Learning to work, Associative Memory AI does not need to be trained. This Associative memory AI does rapid, one-shot learning. It’s a model-free AI.
Associative memory is “white box AI”. It can explain how it have arrived at a certain conclusion. In the past, financial institutions acquired a model-based, vendor-supplied solution for fraud detections.
The different types of AI are complementary.