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

Load balancing in computer networks is one of the most subjects that has got researcher's attention in the last decade. Load balancing will lead to reduce processing time and memory usage that are the most two concerns of the network companies in now days, and they are the most two factors that determine if the approach is worthy applicable or not. There are two kinds of load balancing, distributing jobs among other servers before processing starts and stays at that server to the end of the process is called static load balancing, and moving jobs during processing is called dynamic load balancing. In this research, two algorithms are designed and implemented, the History Usage (HU) algorithm that statically balances the load of a Loaded

planning is among the most significant in the field of robotics research.  As it is linked to finding a safe and efficient route in a cluttered environment for wheeled mobile robots and is considered a significant prerequisite for any such mobile robot project to be a success. This paper proposes the optimal path planning of the wheeled mobile robot with collision avoidance by using an algorithm called grey wolf optimization (GWO) as a method for finding the shortest and safe. The research goals in this study for identify the best path while taking into account the effect of the number of obstacles and design parameters on performance for the algorithm to find the best path. The simulations are run in the MATLAB environment to test the

      In many areas, such as simulation, numerical analysis, computer programming, decision-making, entertainment, and coding, a random number input is required. The pseudo-random number uses its seed value. In this paper, a hybrid method for pseudo number generation is proposed using Linear Feedback Shift Registers (LFSR) and Linear Congruential Generator (LCG). The hybrid method for generating keys is proposed by merging technologies. In each method, a new large in key-space group of numbers were generated separately. Also, a higher level of secrecy is gained such that the internal numbers generated from LFSR are combined with LCG (The adoption of roots in non-linear iteration loops). LCG and LFSR are linear structures and outputs

High frequency (HF) communications have an important role in long distances wireless communications. This frequency band is more important than VHF and UHF, as HF frequencies can cut longer distance with a single hopping. It has a low operation cost because it offers over-the-horizon communications without repeaters, therefore it can be used as a backup for satellite communications in emergency conditions. One of the main problems in HF communications is the prediction of the propagation direction and the frequency of optimum transmission (FOT) that must be used at a certain time. This paper introduces a new technique based on Oblique Ionosonde Station (OIS) to overcome this problem with a low cost and an easier way. This technique uses the

In this paper, we present multiple bit error correction coding scheme based on extended Hamming product code combined with type II HARQ using shared resources for on chip interconnect. The shared resources reduce the hardware complexity of the encoder and decoder compared to the existing three stages iterative decoding method for on chip interconnects. The proposed method of decoding achieves 20% and 28% reduction in area and power consumption respectively, with only small increase in decoder delay compared to the existing three stage iterative decoding scheme for multiple bit error correction. The proposed code also achieves excellent improvement in residual flit error rate and up to 58% of total power consumption compared to the other err

In this paper, an adaptive polynomial compression technique is introduced of hard and soft thresholding of transformed residual image that efficiently exploited both the spatial and frequency domains, where the technique starts by applying the polynomial coding in the spatial domain and then followed by the frequency domain of discrete wavelet transform (DWT) that utilized to decompose the residual image of hard and soft thresholding base. The results showed the improvement of adaptive techniques compared to the traditional polynomial coding technique.

A multidimensional systolic arrays realization of LMS algorithm by a method of mapping regular algorithm onto processor array, are designed. They are based on appropriately selected 1-D systolic array filter that depends on the inner product sum systolic implementation. Various arrays may be derived that exhibit a regular arrangement of the cells (processors) and local interconnection pattern, which are important for VLSI implementation. It reduces latency time and increases the throughput rate in comparison to classical 1-D systolic arrays. The 3-D multilayered array consists of 2-D layers, which are connected with each other only by edges. Such arrays for LMS-based adaptive (FIR) filter may be opposed the fundamental requirements of fa