Design of bidirectional associative memories based on the perceptron training technique

Salih, Ismail; Smith, Stanley H.; Liu, D.

doi:10.1109/iscas.1999.777582

Cited by 6 publications

(4 citation statements)

References 2 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As inspired by the perceptron-learning algorithm, an optimal learning scheme for a class of BAM is advanced [11], which has superior convergence and stability properties. In [12], the synthesis problem of bidirectional associative memories is formulated as a set of linear inequalities that can be solved using the perceptron training algorithm. A multilayer recursive neural network with symmetrical interconnections is introduced in [14] for improved recognition performance under noisy conditions and for increased storage capacity.…”

Section: Related Workmentioning

confidence: 99%

Using Multi-layered Feed-forward Neural Network (MLFNN) Architecture as Bidirectional Associative Memory (BAM) for Function Approximation

Singh¹,

Kumar²

2013

IOSR-JCE

View full text Add to dashboard Cite

Section: Related Workmentioning

confidence: 99%

Using Multi-layered Feed-forward Neural Network (MLFNN) Architecture as Bidirectional Associative Memory (BAM) for Function Approximation

Singh¹,

Kumar²

2013

IOSR-JCE

View full text Add to dashboard Cite

“…A weighted-pattern learning algorithm for BAM is described in [11] by means of global minimization. As inspired by the perceptron-learning algorithm, an optimal learning scheme for a class of BAM is advanced [12]- [13], which has superior convergence and stability properties. In [14], the synthesis problem of bidirectional associative memories is formulated as a set of linear inequalities that can be solved using the perceptron training algorithm.…”

Section: Related Workmentioning

confidence: 99%

Exploring optimal architecture of Multi-layered Feed- forward (MLFNN) as Bidirectional Associative Memory (BAM) for Function Approximation

Singh¹,

Kumar²

2013

IJCA

View full text Add to dashboard Cite

Function approximation is an instance of supervised learning which is one of the most studied topics in machine learning, artificial neural networks, pattern recognition, and statistical curve fitting. In principle, any of the methods studied in these fields can be used in reinforcement learning. Multi-layered feed-forward neural networks (MLFNN) have been extensively used for the purpose of function approximation. Another class of neural networks, BAM, has also been studied and experimented for pattern mapping problems and many variations have been reported in literature. In the present study the application of back propagation algorithm to MLFNN has been proposed in such a way that feed-forward architecture behaves like BAM. Various architectures consisting of fourlayers have been explored in quest of finding the optimal architecture for the example function.

show abstract

“…Due to the nature of the Hebbian learning rule, a Hebb rule trained BAM has a limited storage and mapping capability. Studies (such as [8]- [13]) have been carried out to address the latter problem. In the high-order BAM [8], a high-order nonlinearity is applied to forward and backward information flows to increase the memory capacity and to improve error-correction capability.…”

Section: Introductionmentioning

confidence: 99%

“…As inspired by the perceptronlearning algorithm, an optimal learning scheme for a class of BAM is advanced [12], which has a superior convergence and stability properties. In [13], the synthesis problem of bidirectional associative memories is formulated as a set of linear inequalities that can be solved using the perception-training algorithm. In this paper, three-layer bidirectional symmetrical and asymmetrical associative memories are presented, and a least mean square (LMS) type of delta rule learning algorithm [14] is used for training the networks.…”

Section: Introductionmentioning

confidence: 99%

Three-Layer Symmetrical and Asymmetrical Associative Memories for Image Applications

Kwan

2005 IEEE International Symposium on Circuits and Systems

View full text Add to dashboard Cite

In this paper, three-layer bidirectional symmetrical and asymmetrical associative memories are presented. The networks possess desired structural features of a bidirectional associative memory and a three-layer feedforward neural network and are able to learn and recall binary alphanumeral patterns and gray level images. The network allows two modes of memory recall, namely, recalling by a patternpair from both the input and the output layers, and recalling by single-pattern from either the input layer or the output layer. Simulation results on alpha-numeral and face image types of pattern learning and recall problems are provided.

show abstract

Design of bidirectional associative memories based on the perceptron training technique

Cited by 6 publications

References 2 publications

Using Multi-layered Feed-forward Neural Network (MLFNN) Architecture as Bidirectional Associative Memory (BAM) for Function Approximation

Using Multi-layered Feed-forward Neural Network (MLFNN) Architecture as Bidirectional Associative Memory (BAM) for Function Approximation

Exploring optimal architecture of Multi-layered Feed- forward (MLFNN) as Bidirectional Associative Memory (BAM) for Function Approximation

Three-Layer Symmetrical and Asymmetrical Associative Memories for Image Applications

Contact Info

Product

Resources

About