Search

Recently, many researchers have used nature inspired metaheuristic algorithms due to their ability to perform optimally on complex problems. To solve problems in a simple way, in the recent era bat algorithm has become famous due to its high tendency towards convergence to the global optimum most of the time. But, still the standard bat with random walk has a problem of getting stuck in local minima. In order to solve this problem, this research proposed bat algorithm with levy flight random walk. Then, the proposed Bat with Levy flight algorithm is further hybridized with three different variants of ANN. The proposed BatLFBP is applied to the problem of insulinDNAsequence classification of healthy homosapien. For classification performance, the proposed models such as Bat levy flight Artificial Neural Network (BatLFANN) and Bat levy Flight Back Propagation (BatLFBP) are compared with the other state-of-the-art algorithms like Bat Artificial Neural Network (BatANN), Bat back propagation (BatBP), Bat Gaussian distribution ArtificialNeuralNetwork (BatGDANN). And BatGaussian distribution back propagation (BatGDBP), in-terms of means squared error (MSE) and accuracy. From the perspective of simulations results, it is show that the proposed BatLFANN achieved 99.88153% accuracy with MSE of 0.001185, and BatLFBP achieved 99.834185 accuracy with MSE of 0.001658 on WL5. While on WL10 the proposed BatLFANN achieved 99.89899% accuracy with MSE of 0.00101, and BatLFBP achieved 99.84473% accuracy with MSE of 0.004553. Similarly, on WL15 the proposed BatLFANN achieved 99.82853% accuracy with MSE of 0.001715, and BatLFBP achieved 99.3262% accuracy with MSE of 0.006738 which achieve better accuracy as compared to the other hybrid models. [ABSTRACT FROM AUTHOR]

Recurrent neural network (RNN) has been widely used as a tool in the data classification. This network can be educated with gradient descent back propagation. However, traditional training algorithms have some drawbacks such as slow speed of convergence being not definite to find the global minimum of the error function since gradient descent may get stuck in local minima. As a solution, nature inspired metaheuristic algorithms provide derivative-free solution to optimize complex problems. This paper proposes a new metaheuristic search algorithm called Cuckoo Search (CS) based on Cuckoo bird’s behavior to train Elman recurrent network (ERN) and back propagation Elman recurrent network (BPERN) in achieving fast convergence rate and to avoid local minima problem. The proposed CSERN and CSBPERN algorithms are compared with artificial bee colony using BP algorithm and other hybrid variants algorithms. Specifically, some selected benchmark classification problems are used. The simulation results show that the computational efficiency of ERN and BPERN training process is highly enhanced when coupled with the proposed hybrid method. [ABSTRACT FROM AUTHOR]

Abstract: Recently, the popularity of artificial neural networks (ANN) is increasing since its capacity to model very complex problems in the area of Machine Learning, Data Mining and Pattern Recognition. Improving training efficacy of ANN based algorithm is a dynamic area of research and several papers have been reviewed in the literature. The performance of Multi-layer Perceptrons (MLP) trained with Back Propagation Artificial Neural Network (BP-ANN) method is highly influenced by the size of the datasets and the data-preprocessing techniques used. This work analyses the benefits of using pre-processing datasets using different techniques in-order to improve the ANN convergence. Specifically Min-Max, Z-Score and Decimal Scaling Normalization preprocessing techniques were evaluated. The simulation results show that the computational efficiency of ANN training process is highly enhanced when coupled with different preprocessing techniques. [Copyright &y& Elsevier]

Abstract: Back propagation training algorithm is widely used techniques in artificial neural network and is also very popular optimization task in finding an optimal weight sets during the training process. However, traditional back propagation algorithms have some drawbacks such as getting stuck in local minimum and slow speed of convergence. This research proposed an improved Levenberg Marquardt (LM) based back propagation (BP) trained with Cuckoo search algorithm for fast and improved convergence speed of the hybrid neural networks learning method. The performance of the proposed algorithm is compared with Artificial Bee Colony (ABC) and the other hybridized procedure of its kind. The simulation outcomes show that the proposed algorithm performed better than other algorithm used in this study in term of convergence speed and rate. [Copyright &y& Elsevier]

Training an artificial neural network is an optimization task, since it is desired to find optimal weight sets for a neural network during training process. Traditional training algorithms such as back propagation have some drawbacks such as getting stuck in local minima and slow speed of convergence. This study combines the best features of two algorithms; i.e. Levenberg Marquardt back propagation (LMBP) and Cuckoo Search (CS) for improving the convergence speed of artificial neural networks (ANN) training. The proposed CSLM algorithm is trained on XOR and OR datasets. The experimental results show that the proposed CSLM algorithm has better performance than other similar hybrid variants used in this study. [ABSTRACT FROM AUTHOR]