Analysis of weight initialization routines for conjugate gradient training algorithm with Fletcher-Reeves updates

Masood, Sarfaraz; Doja, M. N.; Chandra, P.

doi:10.1109/ccaa.2016.7813734

Cited by 4 publications

(2 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This algorithm requires more iterations to converge rather than the other CG algorithms; however, the number of computations in each step is significantly reduced as no line search is performed [19]. CGF is an updated version of CG which computes new search direction as the ratio of the norm squared of the current gradient to the norm squared of the previous gradient [26][27][28]. CGP calculates new search direction as the ration of the inner product of the previous change in the gradient with the current gradient divided by the norm squared of the previous gradient [9,28].…”

Section: Resultsmentioning

confidence: 99%

Comparison of Neural Network Training Algorithms for Classification of Heart Diseases

Karim

Niakan

Safdari

2018

IJ-AI

View full text Add to dashboard Cite

<span lang="EN-US">Heart disease is the first cause of death in different countries. Artificial neural network (ANN) technique can be used to predict or classification patients getting a heart disease. There are different training algorithms for ANN. We compared eight neural network training algorithms for classification of heart disease data from UCI repository containing 303 samples. Performance measures of each algorithm containing the speed of training, the number of epochs, accuracy, and mean square error (MSE) were obtained and analyzed. Our results showed that training time for gradient descent algorithms was longer than other training algorithms (8-10 seconds). In contrast, Quasi-Newton algorithms were faster than others (<=0 second). MSE for all algorithms was between 0.117 and 0.228. While there was a significant association between training algorithms and training time (p<0.05), the number of neurons in hidden layer had not any significant effect on the MSE and/or accuracy of the models (p>0.05). Based on our findings, for development an ANN classification model for heart diseases, it is best to use Quasi-Newton training algorithms because of the best speed and accuracy.</span>

show abstract

Section: Resultsmentioning

confidence: 99%

Comparison of Neural Network Training Algorithms for Classification of Heart Diseases

Karim

Niakan

Safdari

2018

IJ-AI

View full text Add to dashboard Cite

show abstract

“…Optimal performance was achieved at the backward process by adapting the learning rate parameter capable of affecting the value of the weighting and reduced the number epoch [16]. Furthermore, when the neural network is trained with conjugate gradient training algorithm having Fletcher-Reeves update, Nguyen-Widrow algorithm converge faster and also generalize better than other weight initialization technique [17]. Nguyen-Widrow weight algorithm was also applied on image compression using multilayer feed-forward artificial neural network.…”

Section: Introductionmentioning

confidence: 99%

Improving self-organizing map with nguyen-widrow initialization lgorithm

Linan

Gerardo

Medina

2019

IJEECS

View full text Add to dashboard Cite

<span lang="EN-US">The quality of cluster result and the learning speed of Self-organizing map (SOM) are dependent on the initialization of weights since the initial values for weight vectors affect the performance of SOM training when applied to clustering. In this paper, improvement of SOM was achieved with the application of the Nguyen-Widrow algorithm to initialize weights. Nguyen-Widrow initialization algorithm is a method for initialization of the weights of neural networks to speed up the training process. Performance of the modified SOM was determined in terms of cluster error rate and the number of iterations to achieve convergence using different datasets and results show that the modified SOM algorithm produces better cluster results and improved training speed compared to traditional SOM.</span>

show abstract

A frequency-domain approach to improve ANNs generalization quality via proper initialization

et al. 2018

View full text Add to dashboard Cite

Analysis of weight initialization routines for conjugate gradient training algorithm with Fletcher-Reeves updates

Cited by 4 publications

References 15 publications

Comparison of Neural Network Training Algorithms for Classification of Heart Diseases

Comparison of Neural Network Training Algorithms for Classification of Heart Diseases

Improving self-organizing map with nguyen-widrow initialization lgorithm

A frequency-domain approach to improve ANNs generalization quality via proper initialization

Contact Info

Product

Resources

About