Long-Term Learning in Soar
Long-term learning in Soar is the process of accumulating procedural knowledge throughout the existence of an intelligent, learning agent. In Soar, such knowledge is in the form of IF-THEN productions, or rules, called “chunks.” The learned chunks are new rules capturing the results of resolving obstacles in the reasoning process. This long-term knowledge is maintained by the system with the expectation that it will be useful during the existence of the agent. Declarative memory is not part of Soar’s long-term memory.
Allen Newell, through his book, “Unified Theories of Cognition” (Newell 1990), proposed many partial theories of cognition, both abstract and human, and offered the Soar architecture as a candidate-unified theory of cognition. He defined a unified theory of cognition as “a single set of mechanisms for all of cognitive behavior.” Three parts of his theory are...
- Doorenbos, R. (1993). Matching 100,000 learned rules. Proceedings of the eleventh national conference on artificial intelligence (pp. 290–296). Menlo Park, CA: AAAI Press.Google Scholar
- Kennedy, W. G., & De Jong, K. A. (2003) Characteristics of long-term learning in soar and its application to the utility problem. Proceedings of the twentieth international conference on machine learning (pp. 337–344). Menlo Park, CA: AAAI Press.Google Scholar
- Kennedy, W. G., & Trafton, J. G. (2007). Long-term symbolic learning. Cognitive Systems Research, 8, 237–247.Google Scholar
- Laird, J. E. (2008) Extending the soar cognitive architecture. In Artificial general intelligence 2008: Proceedings of the first AIG conference. Memphis, TN: ISO Press.Google Scholar
- Minton, S. (1990). Quantitative results concerning the utility of explanation-based learning. Artificial Intelligence, 42, 363–391.Google Scholar
- Newell, A. (1990). Unified theories of cognition. Cambridge, MA: Harvard University Press.Google Scholar
- Tambe, M., Newell, A., & Rosenbloom, P. S. (1990). The problem of expensive chunks and its solution by restricting expressiveness. Machine Learning, 5, 299–348.Google Scholar