On fast retrieval of relational experiences

Various intelligent systems are needed for cyber-physical systems. Such intelligent systems need to learn from the human intelligence about concept abstraction and analogical thinking in order to resolve complex issues using past experiences. The algorithms for abstraction and analogies are based on quick memory recall with clever information coding and processing. The quick and accurate memory recall is based on the fact that the memory mostly records the relations among the constituents of the stimulating signals, rather than the constituents themselves. Relational memories can be stored in the form of networks of neuron clusters capable of resonating to particular signal sequences. However, similarity testing for such network representations is difficult. We suggest that linear dynamic systems that relate the system matrix and the output time function can be used as a conversion mechanism between the network matrix and the temporal representations of the signals. This leads to algorithms for relational similarity testing and concept abstraction. Transient behavior based selection rules in ordinal optimization is important in achieving quickness in our development.