Hopfield model

(redirected from Hopfield net)
Also found in: Wikipedia.

Hopfield model

Mentioned in ?
References in periodicals archive ?
Each way corresponds to a fixed point of the Hopfield net [27].
As mentioned previously, a Hopfield net can serve as an associative memory.
Based on these characteristics of electronic meetings and our experience with several useful AI and information retrieval-based algorithms, we selected techniques from the following three areas: automatic indexing, cluster analysis, and Hopfield net classification.
And finally, resurgent neural network computing such as the Hopfield net algorithm [23] has been shown to be useful in performing memory association and classification [3].
Finally, we used the Hopfield net algorithm to group similar terms, i.
Hopfield Net Classification: when pairwise similarities are obtained between all term pairs, a hierarchical agglomerative cluster generation process is often adopted [19].
After a review of numerous networks [14, 15] and based on our experience and experimentation with several major networks [3], the Hopfield net [10], which was introduced as a neural network that can be used as a content-addressable memory, appeared to be a natural candidate for EBS ideas classification.
The parallel relaxation and convergence properties of Hopfield net activation and its novel characteristics for memory association and classification led us to experiment with this newer approach for EBS comment classification.
A sketch of the Hopfield net concept classification procedure follows:
Recall that pairs of units in a Hopfield net are connected by symmetric weights.
In contrast to the parallel relaxation method used by Hopfield nets, backpropagation networks perform a simpler computation.
Various ANN topologies have been proposed to data such as Hopfield nets, Hamming nets, Carpenter/ Grossberg classifiers, perceptrons, multilayer perceptrons, and Kohonen self organizing maps (Haykin, 1994).