If the less significant terms are also included to the concept space, the training of the Hopfield network may not converge because the concept space may continuously expand until it includes a large collection of relevant and irrelevant terms.
The operation of the Hopfield network has two phases, the storage phase and the retrieval phase. In the storage phase, the synaptic weight from neuron j to neuron k, Wjk,are generated. In the retrieval phase, activation of neurons will be applied until convergence.
The Hopfield network can be considered as a nonlinear associate memory or content-addressable memory. The objective of Hopfield network is to retrieve a pattern (concept space) stored in memory in response to the presentation of an incomplete or noisy version of that pattern (Haykin, 1994). The content-addressable memory is error-correcting that overrides inconsistent information in the cues presented to it.