,[11 ii~(lfi(lal Ii(ural nt’twork et[uip[)ed with a proper Ifar[iill: PLIICcali act iiS alI associative memory, patIcr]l cliissifitr ailcl feature extractor. Of course, the SLI(’(’<’SS of’ Sllcll ail a.igorithm crucially depends on [ IL(,cal)iil)ility of” learning the connection weights in 1he l~e[iral IIet\vork. Such learning rules are usually Illodell(d I)y no[lliuear dynamical systems, evolv...