The capacity of the Hopfield associative memory
IEEE Transactions on Information Theory
Bidirectional associative memories
IEEE Transactions on Systems, Man and Cybernetics
A Neuron-Weighted Learning Algorithm and its Hardware Implementation in Associative Memories
IEEE Transactions on Computers
Stability analysis of neural-network interconnected systems
IEEE Transactions on Neural Networks
Associative Memory Design Using Support Vector Machines
IEEE Transactions on Neural Networks
| Hi-index | 0.00 |
A neural model for an intraconnected parallel bidirectional associative memory (PBAM) is proposed. In PBAM, the N-bit and P-bit long binary vectors in two associated sets X and Yare concatenated together to form a set of M compound vectors, and with N=P, it is shown that for L=2N, an improvement of 41.40% in signal to noise ratio and about 100% in storage capacity is achieved as compared to sequential BAM. In this model both interfiled and intrafield connections are present in each neuron field. In the recall processes of PBAM, the intralayer feedback processes run in parallel, whereas in other bidirectional and intraconnected memory models they run in sequential [8, 12]. PBAM being a fully connected model, eliminates the continuity requirements, improves signal to noise ratio, storage capacity, reliability, and allows energy changes resulting from the state changes of neurons in both fields Fx and Fy which help improve the convergence. The performance of PBAM is verified by means of numerical examples.