Self-managing associative memory for dynamic acquisition of expertise in high-level domains
Jacob Beal · 2009
Self-organizing maps can be used to implement an associative memory for an intelligent system that dynamically learns about new high-level domains over time. SOMs are an attractive option for imple-menting associative memory: they are fast, easily parallelized, and digest a stream of incoming data into a topographically organized collection of mod-els where more frequent classes of data are repre-sented by higher-resolution collections of models. Typically, the distribution of models in an SOM, once developed, remains fairly stable, but develop-ing expertise in a new high-level domain requires altering the allocation of models. We use a mixture of analysis and empirical studies to characterize the behavior of SOMs for high-level associative mem-ory, finding that new high-resolution collections of models develop quickly. High-resolution areas of the SOM decay rapidly unless actively refreshed, but in a large SOM, the ratio between growth rate and decay rate may be high enough to support both fast learning and long-term memory. 1