next up previous contents
Next: Automatic Generation of Up: Future Research Previous: History Keeping

Learning in SiFA Based Design Systems

  Future research on learning using the SiFA paradigm will have two positive results. The first is that adding learning capabilities to SiFAs will drastically improve the performance of SiFA based design systems. Secondly, the SiFA paradigm will act as an exceptional tool to investigate learning in multi-agent design systems in general, not just those with SiFAs. This work will also contribute to distributed learning research.

In the SiFA framework, there are many elements that support learning. These are:

The things that might be learned in the SiFA framework are:

Learning should improve the performance of the design system. The aspects of performance that can be improved in the SiFA paradigm are:

Learning in a system with agents can have two distinct results. The first would be to improve the local performance of agents. The second is an improvement of the overall system through organizational learning where the newly acquired knowledge of an agents does not directly contribute to its local goals but has the effect of improving the overall performance.



Ilan Berker
Thu Apr 27 16:25:38 EDT 1995