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

Three cohesionless free flowing materials of different density were mixed in an air fluidized bed to study the mixing process by calculating performance of mixing index according to Rose equation (1959) and to study the effect of four variables (air velocity, mixing time, particle size of trace component and concentration of trace component) on the mixing index and as well as on mixing performance. It was found that mixing index increases with increasing the air velocity, mixing time and concentration of trace component until the optimum value. Mixing index depends on the magnitude of difference in particle size The first set of experiments (salt then sand then cast iron) give higher mixing index and better performance of mixing than the

            Time is very important in educational institutions. It is also one of our contemporary problem ‚as time is a clear – cut and limited factor‚ it demands that administrators should monitor it by administering and monitoring the principles of time.

Hence‚ the researcher attempts to identify the skills of administrating time and the reasons that cause the waste of time of the Heads of Departments at university of Baghdad.

Significance  of the research:

Time is very important to all educational administrators and one of them is the institutions of Higher education. One of the

Background: The ideal force-delivery system must: provide optimal tooth moving forces that elicit the desired effects, be comfortable and hygienic for the patient, require minimal operator manipulation and patient cooperation and provide rapid tooth movement with minimal mobility during orthodontic therapy, the elastomeric chains have the greatest potential to fulfill these requirements. Materials and Methods: This in vitro study was designed to determine the effect of three different mechanisms for canine retraction : (6-3 , 6-5-3 and chain loop ) on the load relaxation behavior of three types of elastomeric chains : (maximum clear , maximum silver and extreme silver) from the same company (Ortho Technology company) with two different bran

In this paper, the memorization capability of a multilayer interpolative neural network is exploited to estimate a mobile position based on three angles of arrival. The neural network is trained with ideal angles-position patterns distributed uniformly throughout the region. This approach is compared with two other analytical methods, the average-position method which relies on finding the average position of the vertices of the uncertainty triangular region and the optimal position method which relies on finding the nearest ideal angles-position pattern to the measured angles. Simulation results based on estimations of the mobile position of particles moving along a nonlinear path show that the interpolative neural network approach outperf