The power of Saffron’s Associative Memory-based approach rests in its simplicity, which is inspired by Natural Intelligence. It’s about understanding your Experience
The concept is simple: We see things and learn how they are related when we see them together (or in sequence). We remember the coincidences. We also reason from this experience of how things are connected. New situations remind us of prior situations, recalling what other things, actions or outcomes are likely connected, while imagining what’s missing, or noticing what’s new and different from what we already know.
As in human experience, computers should also capture such empirical knowledge from direct observation of the real world.
Associative Memory technology works the exact same way we store and use information in our own brains. And if technology is to assist us in transforming data into intelligence, and intelligence into actionable Experience, then it needs to also be more natural. It needs to be more brain-like. A more revolutionary approach, inspired by natural systems, is required to address the growing problem of data management, so we can transform the data itself, and make it more useful to us.
SaffronMemoryBase, the engine inside the Saffron Natural Intelligence Platform, is an entirely new kind of store; not for data as in a database, but for information, knowledge – even wisdom – in a MemoryBase.
It’s about your Experience. Because that’s what it stores.
As We May Think
It’s not a new concept.
Aristotle believed that knowledge was defined by associationism, the collection and recollection of mental connections. Aristotle also described “practical knowledge”. In contrast to general rules, practical knowledge is used to understand and act in each particular case.
Beyond abstract rules and statistics, real-world practitioners (whether they are doctors, bankers, lawyers, or police) reason from past experiences with similar cases. The discipline of Psychology is founded on associationism, seeing it as the elemental study of how sensations, ideas, and actions relate to each other. The foundations of modern neuroscience likewise are based on “connectionism,” and one of our most well-developed understandings of neural functions is based on memory. Whether of motor memories for how to act, or semantic memories of how to think, we practice and reason from memory.
“The human mind…operates by association. Selection by association, rather than indexing, may yet be mechanized.”
- Vandemar Bush, 1945
Although the human brain is more complex than SaffronMemoryBase, we’re always looking for new ways to make computers work more like our brains do, to better assist us.
Dr. Vannevar Bush, sometimes called the father of modern American science, was notably the first to articulate the idea of a human-like associative memory for computers in his article “As We May Think” published July 1945.
As Bush suggested, “The human mind…operates by association. Selection by association, rather than indexing, may yet be mechanized.” We still rely on database indexing and search engine indexing. Document hyper-linking, which became the basis of the World Wide Web, has its intellectual roots in Bush’s idea of computational association, but he wanted computers to be so much more. Bush imagined a personal assistant that would read everything and connect all the elements within and across documents, not just connect the documents. Sixty years later computing is still fixed on indexing rather than associating.
Well, it was, until Saffron became the first software company to commercialize an Associative Memory-based, highly scalable solution.
A Simple Idea
Reasoning by memory is a simple idea. The representation of memory is also simple, at least in principle. Associations represent coincidences of how two things co-occur in data, either at the same time or in sequence. Coincidences are stored in matrices. The coincidence-count between two things is stored in the matrix cell for the row and column of the things. In other words, an associative memory is an associative matrix, or coincidence matrix.
Think about it this way: Imagine a network of memories – think millions of them. As if you had a million personal assistants. Then, you give each of them millions of documents to read. Finally, imagine they can quickly read and remember … everything.
“The intensive use of memory to recall specific
episodes from the past – rather than rules – should
be the foundation of machine reasoning.”
– Stanfill & Waltz, Toward Memory-based Reasoning, 1986
In SaffronMemoryBase, each memory concentrates on a single person, place, thing, event, action or outcome. Each memory observes, stores and recalls every association among people, places things and events in great detail and in context. Think millions of attributes within each memory. And best of all, these memories never forget – they have total recall.
This was the vision of Vannevar Bush. He imagined a personal assistant to read and remember everything.
Bush’s vision has not been available before Saffron because of the matrix scaling problem. Saffron has solved the scaling of associative matrices to such numbers and sizes. But the fundamental idea of associative memory is simple, and this simplicity is its power.
Saffron does with your Experience what could not be done with data analytics before now:
It gives you a virtual army of personal assistants for everything you do.