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 Hermite interpolation method is used for solving linear and non-linear second order singular multi point boundary value problems with nonlocal condition. The approximate solution is found in the form of a rapidly convergent polynomial. We discuss behavior of the solution in the neighborhood of the singularity point which appears to perform satisfactorily for singular problems. The examples to demonstrate the applicability and efficiency of the method have been given.

One of the challenges faced by traditional intrusion detection systems based on machine learning or deep learning is instability when dealing with unbalanced network traffic, leading to failure in detecting certain attacks (minority classifications). Additionally, they struggle with multi-stage attacks, resulting in an increase in false alarms. This paper presents a hierarchical intrusion detection system supported by a Prediction Verification Layer (PVL) and a Feature Inheritance Mechanism (FIM). Where PVL contributes to documenting the system’s final decision and increasing sensitivity to minority attacks, FIM also helps in inheriting features from previous layers and correcting errors as much as possible. Additionally, it allows for ad

In this study, nanocomposites have been prepared by adding
multiwall carbon nanotubes (MWCNTs) with weight ratios (0, 2, 3,
4, 5) wt% to epoxy resin. The samples were prepared by hand lay-up
method. Influence of an applied load before and after immersion in
sodium hydroxide (NaOH) of normality (0.3N) for (15 days) at
laboratory temperature on wear rate of Ep/MWCNTs
nanocomposites was studied. The results showed that wear rate
increases with increasing the applied load for the as prepared and
immersed samples and after immersion. It was also found that epoxy
resin reinforced with MWCNTs has wear rate less than neat epoxy.
The sample (Ep + 5wt% of MWCNTs) has lower wear rate. The
immersion effect in base so

Credit risk assessment has become an important topic in financial risk administration. Fuzzy clustering analysis has been applied in credit scoring. Gustafson-Kessel (GK) algorithm has been utilised to cluster creditworthy customers as against non-creditworthy ones. A good clustering analysis implemented by good Initial Centres of clusters should be selected. To overcome this problem of Gustafson-Kessel (GK) algorithm, we proposed a modified version of Kohonen Network (KN) algorithm to select the initial centres. Utilising similar degree between points to get similarity density, and then by means of maximum density points selecting; the modified Kohonen Network method generate clustering initial centres to get more reasonable clustering res

Wireless sensor networks (WSNs) represent one of the key technologies in internet of things (IoTs) networks. Since WSNs have finite energy sources, there is ongoing research work to develop new strategies for minimizing power consumption or enhancing traditional techniques. In this paper, a novel Gaussian mixture models (GMMs) algorithm is proposed for mobile wireless sensor networks (MWSNs) for energy saving. Performance evaluation of the clustering process with the GMM algorithm shows a remarkable energy saving in the network of up to 92%. In addition, a comparison with another clustering strategy that uses the K-means algorithm has been made, and the developed method has outperformed K-means with superior performance, saving ener

This paper presents a hybrid genetic algorithm (hGA) for optimizing the maximum likelihood function ln(L(phi(1),theta(1)))of the mixed model ARMA(1,1). The presented hybrid genetic algorithm (hGA) couples two processes: the canonical genetic algorithm (cGA) composed of three main steps: selection, local recombination and mutation, with the local search algorithm represent by steepest descent algorithm (sDA) which is defined by three basic parameters: frequency, probability, and number of local search iterations. The experimental design is based on simulating the cGA, hGA, and sDA algorithms with different values of model parameters, and sample size(n). The study contains comparison among these algorithms depending on MSE value. One can conc