An Associative Memory is an information store that organizes data in terms of relationships. In essence, an associative memory is “correlation coefficient analysis on steroids”.
The Associative Memory module learns quickly and can be extended on-the-fly as new data types become available. It is particularly useful for tasks that require pattern matching when some data may be missing or even incorrect.
Unlike techniques that fail when data are missing or incorrect, the performance of the associative memory degrades gracefully, just as human performance does in analogous situations.
The Associative Memory module provides capabilities that augment VELOX.engine, our OODA Loop (Observe / Orient / Decide / Act) cognitive engines. It provides another technique for making good decisions in dynamic environments.
In a seminal 1945 essay titled As We May Think, Vannevar Bush argued that computers in the future would be associative. Bush, a leading figure in the Manhattan Project, understood that significant aspects of learning involve forming associations. We connect events, people, places and facts together and use these associations in all sorts of ways. For example, countless people remember exactly where they were and what they were doing on 9/11.