Behavioral Fault Model for Neural Networks

Ahmadi, Ahmad; Fakhraie, S.M.; Lucas, Caro

doi:10.1109/iccet.2009.201

Cited by 5 publications

(5 citation statements)

References 5 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In our MPE formulae, the complexity of the direct calcula- 1 Since the number M of nodes is usually less than the number N of training samples, the complexity is equal to O(M × N). 1 Although the computational cost of Tr (29) and (36), becomes non-diagonal in the eigen domain of H φ .…”

Section: Complexity Of Mpe Formulaementioning

confidence: 99%

“…1 Although the computational cost of Tr (29) and (36), becomes non-diagonal in the eigen domain of H φ . Hence, the complexity is equal to O(M × N + M 2 ).…”

Section: Complexity Of Mpe Formulaementioning

confidence: 99%

“…5 and In the classical results on generalization errors [1], [31], [32] for the fault-free case, there are also some differences between the true test error and the estimated test error. In our MPE results for the fault case, the differences come from the approximation and assumptions in the analysis.…”

Section: ) Regressionmentioning

confidence: 99%

“…When some weights are deviated from their trained values, or some hidden nodes of a trained network are out of work, the degradation in the performance of the trained network should not be so large. In fact, many neural network pioneers [1], [6], [11], [34], [37], [38], [42] have investigated the property of network fault and proposed a number of methods to improve the fault tolerance.…”

mentioning

confidence: 99%

See 3 more Smart Citations

On the Selection of Weight Decay Parameter for Faulty Networks

Leung

Wang

Sum

2010

IEEE Trans. Neural Netw.

View full text Add to dashboard Cite

Abstract-The weight-decay technique is an effective approach to handle overfitting and weight fault. For fault-free networks, without an appropriate value of decay parameter, the trained network is either overfitted or underfitted. However, many existing results on the selection of decay parameter focus on fault-free networks only. It is well known that the weight-decay method can also suppress the effect of weight fault. For the faulty case, using a test set to select the decay parameter is not practice because there are huge number of possible faulty networks for a trained network. This paper develops two mean prediction error (MPE) formulae for predicting the performance of faulty radial basis function (RBF) networks. Two fault models, multiplicative weight noise and open weight fault, are considered. Our MPE formulae involve the training error and trained weights only. Besides, in our method, we do not need to generate a huge number of faulty networks to measure the test error for the fault situation. The MPE formulae allow us to select appropriate values of decay parameter for faulty networks. Our experiments showed that, although there are small differences between the true test errors (from the test set) and the MPE values, the MPE formulae can accurately locate the appropriate value of the decay parameter for minimizing the true test error of faulty networks.

show abstract

Section: Complexity Of Mpe Formulaementioning

confidence: 99%

“…1 Although the computational cost of Tr (29) and (36), becomes non-diagonal in the eigen domain of H φ . Hence, the complexity is equal to O(M × N + M 2 ).…”

Section: Complexity Of Mpe Formulaementioning

confidence: 99%

Section: ) Regressionmentioning

confidence: 99%

mentioning

confidence: 99%

See 2 more Smart Citations

On the Selection of Weight Decay Parameter for Faulty Networks

Leung

Wang

Sum

2010

IEEE Trans. Neural Netw.

View full text Add to dashboard Cite

show abstract

“…Ahmadi et al proposed a behavioral fault model in [8] for neural networks. Authors took advantage of inherent properties of neural network in error masking and made the fault model in an abstract level corresponding to the neural graph.…”

Section: Introductionmentioning

confidence: 99%

Behavioral Modeling for Public Safety Communication in Wireless Mesh Networking

Wei

Zhang

2010

2010 International Conference on Computational Intelligence and Software Engineering

View full text Add to dashboard Cite

Accompanying dramatic advances in wireless Mesh networking, the demand for advanced mechanisms to employ wireless Mesh networks in public safety continues to grow. However, this special field communication brings new security challenges. In this paper, we import the expectancy-value theory in cognitive psychology to behavioral evaluation of node and propose a behavioral model named NODE EYE for public safety communication in wireless Mesh networks and exercise the model to reason a variation of station-to-station key agreement protocol. We contend that our model facilitates reasoning about protocols designed for public safety by integrating the behavioral trust and cryptographic aspects of communication in the special field. We illustrate our concept by introducing the key agreement to support the special communication cases associated with the public safety.

show abstract