E. Ruppin and M. Usher. An attractor neural network model of semantic fact retrieval. Network, 1(3):325, 1990.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....space. For instance, one approach here is for representations of entities in a single inheritance hierarchy to be computed with respect to their position in the hierarchy according to some formula. Attributes can be attached to isa related entities (inheritance) through an attractor neural network (Usher and Ruppin, 1990). To obtain meaningful distributed representations, one can either code semantic features of an entity as microfeatures or let the processing task (e.g. associating semantically similar entities) decide what representation is appropriate for a specific entity (Miikkulainen and Dyer, 1988) In the ....

Usher, M. and Ruppin, E. (1990). An attractor neural network model of semantic fact retrieval. In Proceedings of IJCNN.

....space. For instance, one approach here is for representations of entities in a single inheritance hierarchy to be computed with respect to their position in the hierarchy according to some formula. Attributes can be attached to isa related entities (inheritance) through an attractor neural network [16]. To obtain meaningful distributed representations, one can either code semantic features of an entity as microfeatures or let the processing task (e.g. associating semantically similar entities) decide what representation is appropriate for a specific entity [9] In the latter case, the process ....

M. Usher and E. Ruppin. An attractor neural network model of semantic fact retrieval. In Proceedings of IJCNN, 1990.

....performing error correction, they can perform content addressable memory retrieval. Indeed, ANN models of psychological data concerning specific aspects of memory retrieval have been presented; e.g. of high speed scanning experiments of the Sternberg type [ASU90] and of semantic memory queries [RU90]. Previous classical mathematical models of memory retrieval (reviewed in [GS84] have shown a remarkable ability to fit experimental data. However, these models entail the existence of numerous parameters bearing arbitrarily assigned high level cognitive interpretation . Moreover, the broad ....

.... differences between them (two process theories) Kin70] Indeed, in the model presented, both Recall and Recognition are performed in the same network, sharing the same dynamics; the dynamic behavior of an ANN can be viewed as performing a parallel, mutual exclusive search in the phase space [RU90]. Yet, during a Recognition assignment, the read out of this similar process is different than in Recall, since familiarity is then examined. For a similar observation regarding feed forward networks see [YRD79] 3 The modeling of experimental data. In this section we describe how various ....

[Article contains additional citation context not shown here]

E. Ruppin and M. Usher. An attractor neural network model of semantic fact retrieval. Network, 1(3):325, 1990.

....an associative like manner if it is cued by an input pattern that is sufficiently similar to it. The stored patterns are therefore called attractors. Such associative neural models have been shown to provide interesting insights both to normal human memory function and to its degradation (e.g. [14, 15, 16, 17, 18]) These models have also received biological support from electrophysiological studies showing delayed, sustained activity in the brain during memory related tasks [19, 20] In a recent study [21] we have modeled voltage dependence of NMDA receptors as a learning threshold. The functional ....

E. Ruppin and M. Usher. An attractor neural network model of semantic fact retrieval. Network, 1(3):325, 1990.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC