A three-stage learning algorithm for deep multilayer perceptron (DMLP) with effective weight initialisation based on sparse auto-encoder is proposed in this paper, which aims to overcome difficulties in training deep neural networks with limited training data in high-dimensional feature space. At the first stage, unsupervised learning is adopted using sparse auto-encoder to obtain the initial weights of the feature extraction layers of the DMLP. At the second stage, error back-propagation is used to train the DMLP by fixing the weights obtained at the first stage for its feature extraction layers. At the third stage, all the weights of the DMLP obtained at the second stage are refined by error back-propagation. Network structures an

 In this paper, the human robotic leg which can be represented mathematically by single input-single output (SISO) nonlinear differential model with one degree of freedom, is analyzed and then a simple hybrid neural fuzzy controller is designed to improve the performance of this human robotic leg model. This controller consists from SISO fuzzy proportional derivative (FPD) controller with nine rules summing with single node neural integral derivative (NID) controller with nonlinear function. The Matlab simulation results for nonlinear robotic leg model with the suggested controller showed that the efficiency of this controller when compared with the results of the leg model that is controlled by PI+2D, PD+NID, and F

Channel estimation (CE) is essential for wireless links but becomes progressively onerous as Fifth Generation (5G) Multi-Input Multi-Output (MIMO) systems and extensive fading expand the search space and increase latency. This study redefines CE support as the process of learning to deduce channel type and signal-tonoise ratio (SNR) directly from per-tone Orthogonal Frequency-Division Multiplexing (OFDM) observations,with blind channel state information (CSI). We trained a dual deep model that combined Convolutional Neural Networks (CNNs) with Bidirectional Recurrent Neural Networks (BRNNs). We used a lookup table (LUT) label for channel type (class indices instead of per-tap values) and ordinal supervision for SNR (0–20 dB,5-dB steps). T

In this paper, some commonly used hierarchical cluster techniques have been compared. A comparison was made between the agglomerative hierarchical clustering technique and the k-means technique, which includes the k-mean technique, the variant K-means technique, and the bisecting K-means, although the hierarchical cluster technique is considered to be one of the best clustering methods. It has a limited usage due to the time complexity. The results, which are calculated based on the analysis of the characteristics of the cluster algorithms and the nature of the data, showed that the bisecting K-means technique is the best compared to the rest of the other methods used.

Whenever, the Internet of Things (IoT) applications and devices increased, the capability of the its access frequently stressed. That can lead a significant bottleneck problem for network performance in different layers of an end point to end point (P2P) communication route. So, an appropriate characteristic (i.e., classification) of the time changing traffic prediction has been used to solve this issue. Nevertheless, stills remain at great an open defy. Due to of the most of the presenting solutions depend on machine learning (ML) methods, that though give high calculation cost, where they are not taking into account the fine-accurately flow classification of the IoT devices is needed. Therefore, this paper presents a new model bas

This paper proposes a new structure of the hybrid neural controller based on the identification model for nonlinear systems. The goal of this work is to employ the structure of the Modified Elman Neural Network (MENN) model into the NARMA-L2 structure instead of Multi-Layer Perceptron (MLP) model in order to construct a new hybrid neural structure that can be used as an identifier model and a nonlinear controller for the SISO linear or nonlinear systems. Weight parameters of the hybrid neural structure with its serial-parallel configuration are adapted by using the Back propagation learning algorithm. The ability of the proposed hybrid neural structure for nonlinear system has achieved a fast learning with minimum number

To date, comprehensive reviews and discussions of the strengths and limitations of Remote Sensing (RS) standalone and combination approaches, and Deep Learning (DL)-based RS datasets in archaeology have been limited. The objective of this paper is, therefore, to review and critically discuss existing studies that have applied these advanced approaches in archaeology, with a specific focus on digital preservation and object detection. RS standalone approaches including range-based and image-based modelling (e.g., laser scanning and SfM photogrammetry) have several disadvantages in terms of spatial resolution, penetrations, textures, colours, and accuracy. These limitations have led some archaeological studies to fuse/integrate multip

In this paper, a robust invisible watermarking system for digital video encoded by MPEG-4 is presented. The proposed scheme provides watermark hidden by embedding a secret message (watermark) in the sprite area allocated in reference frame (I-frame). The proposed system consists of two main units: (i) Embedding unit and (ii) Extraction unit. In the embedding unit, the system allocates the sprite blocks using motion compensation information. The allocated sprite area in each I–frame is used as hosting area for embedding watermark data. In the extraction unit, the system extracts the watermark data in order to check authentication and ownership of the video. The watermark data embedding method is Blocks average modulation applied on RGB dom

The multiple linear regression model is an important regression model that has attracted many researchers in different fields including applied mathematics, business, medicine, and social sciences , Linear regression models involving a large number of independent variables are poorly performing due to large variation and lead to inaccurate conclusions , One of the most important problems in the regression analysis is the multicollinearity Problem, which is considered one of the most important problems that has become known to many researchers  , As well as their effects on the multiple linear regression model, In addition to multicollinearity, the problem of outliers in data is one of the difficulties in constructing the reg