|Wednesday, January 05, 2005|
|Ai releases HAL3000|
|Savyon, Israel -- January 5, 2005 - Artificial Intelligence Research (Ai) |
(http://www.a-i.com) announced today the release of a new version of its Learning Algorithm: HAL3000.
In 2002 thru 2004, the three major versions publicly released by Ai (known as HAL, HAL2003 and HAL2004 respectively) all implemented algorithms for learning how context effects response selection and algorithms for generalization and classification via the formation of a rule system. As the series progressed, the role of the rule system grew stronger, sometimes to the point of taking over context considerations.
Like HAL2004, HAL3000 is implemented in Java. It provides for a cleaner break between context processing and rule formation. HAL3000 differs from all previous versions in its context model, which it considers a complete and true description of all context dependencies for its total history.
Unlike previous models, which were statistical and used probabilities, HAL3000 has one and only "explanation" for each correct response in the entire history. Its rule system does not compete with the context model, but completes it.
Balancing Rules with Context
Like in HAL2004, rules (with corresponding generalization and classes) are used to process new input/context combinations and propose newly constructed responses. HAL3000 either knows exactly what do say, or alternatively uses a set of rules to "invent" a new response.
Artificial Intelligence Research (Ai) is a groundbreaking research project, focusing on creating genuine Artificial Intelligence – the technology that enables machines to converse with humans in natural language. Ai's research facility is located in Savyon, Israel.