The aim of this paper, is to discuss several high performance training algorithms fall into two main categories. The first category uses heuristic techniques, which were developed from an analysis of the performance of the standard gradient descent algorithm. The second category of fast algorithms uses standard numerical optimization techniques such as: quasi-Newton . Other aim is to solve the drawbacks related with these training algorithms and propose an efficient training algorithm for FFNN

The deep learning algorithm has recently achieved a lot of success, especially in the field of computer vision. This research aims to describe the classification method applied to the dataset of multiple types of images (Synthetic Aperture Radar (SAR) images and non-SAR images). In such a classification, transfer learning was used followed by fine-tuning methods. Besides, pre-trained architectures were used on the known image database ImageNet. The model VGG16 was indeed used as a feature extractor and a new classifier was trained based on extracted features.The input data mainly focused on the dataset consist of five classes including the SAR images class (houses) and the non-SAR images classes (Cats, Dogs, Horses, and Humans). The Conv

In this paper, we derive and prove the stability bounds of the momentum coefficient µ and the learning rate ? of the back propagation updating rule in Artificial Neural Networks .The theoretical upper bound of learning rate ? is derived and its practical approximation is obtained

One of the most interested problems that recently attracts many research investigations in Protein-protein interactions (PPI) networks is complex detection problem. Detecting natural divisions in such complex networks is proved to be extremely NP-hard problem wherein, recently, the field of Evolutionary Algorithms (EAs) reveals positive results. The contribution of this work is to introduce a heuristic operator, called protein-complex attraction and repulsion, which is especially tailored for the complex detection problem and to enable the EA to improve its detection ability. The proposed heuristic operator is designed to fine-grain the structure of a complex by dividing it into two more complexes, each being distinguished with a core pr

The efficiency evaluation of the railway lines performance is done through a set of indicators and criteria, the most important are transport density, the productivity of enrollee, passenger vehicle production, the productivity of freight wagon, and the productivity of locomotives. This study includes an attempt to calculate the most important of these indicators which transport density index from productivity during the four indicators, using artificial neural network technology. Two neural networks software are used in this study, (Simulnet) and (Neuframe), the results of second program has been adopted. Training results and test to the neural network data used in the study, which are obtained from the international in

The evolution in the field of Artificial Intelligent (AI) with its training algorithms make AI very important in different aspect of the life. The prediction problem of behavior of dynamical control system is one of the most important issue that the AI can be employed to solve it. In this paper, a Convolutional Multi-Spike Neural Network (CMSNN) is proposed as smart system to predict the response of nonlinear dynamical systems. The proposed structure mixed the advantages of Convolutional Neural Network (CNN) with Multi -Spike Neural Network (MSNN) to generate the smart structure. The CMSNN has the capability of training weights based on a proposed training algorithm. The simulation results demonstrated that the proposed

Several stress-strain models were used to predict the strengths of steel fiber reinforced concrete, which are distinctive of the material. However, insufficient research has been done on the influence of hybrid fiber combinations (comprising two or more distinct fibers) on the characteristics of concrete. For this reason, the researchers conducted an experimental program to determine the stress-strain relationship of 30 concrete samples reinforced with two distinct fibers (a hybrid of polyvinyl alcohol and steel fibers), with compressive strengths ranging from 40 to 120 MPa. A total of 80% of the experimental results were used to develop a new empirical stress-strain model, which was accomplished through the application of the parti

This paper presents a modified training method for Recurrent Neural Networks. This method depends on the Non linear Auto Regressive (NARX) model with Modified Wavelet Function as activation function (MSLOG) in the hidden layer. The modified model is known as Modified Recurrent Neural (MRN). It is used for identification Forward dynamics of four Degrees of Freedom (4-DOF) Selective Compliance Assembly Robot Arm (SCARA) manipulator robot. This model is also used in the design of Direct Inverse Control (DIC). This method is compared with Recurrent Neural Networks that used Sigmoid activation function (RS) in the hidden layer and Recurrent Neural Networks with Wavelet activation function (RW). Simulation results shows that the MRN model is bett

The aim of this paper is to approximate multidimensional functions f∈C(R^s) by developing a new type of Feedforward neural networks (FFNS) which we called it Greedy ridge function neural networks (GRGFNNS). Also, we introduce a modification to the greedy algorithm which is used to train the greedy ridge function neural networks. An error bound are introduced in Sobolov space. Finally, a comparison was made between the three algorithms (modified greedy algorithm, Backpropagation algorithm and the result in [1]).