Abstract In this study thin-layer drying of bergamot was modelled using artificial neural network. An experimental dryer was used. Thin-layer of bergamot slices at five air temperatures (40, 50, 60, 70 & 80 ºC), one thickness (6 mm) and three air velocities (0.5, 1 & 2 m/s) were artificially dried. Initial moisture content (M.C.) during all experiments was between 5.2 to... In this study thin-layer drying of bergamot was modelled using artificial neural network. An experimental dryer was used. Thin-layer of bergamot slices at five air temperatures (40, 50, 60, 70 & 80 ºC), one thickness (6 mm) and three air velocities (0.5, 1 & 2 m/s) were artificially dried. Initial moisture content (M.C.) during all experiments was between 5.2 to 5.8 (g.g) (d.b.). Mass of samples were recorded and saved every 5 sec. using a digital balance connected to a PC. MLP with momentum and levenberg-marquardt (LM) were used to train the ANNS. In order to develop ANN’s models, temperatures, air velocity and time are used as input vectors and moisture ration as the output. Results showed a 3-8-1 topology for thickness of 6 mm, with LM algorithm and TANSIG activation function was able to predict moisture ratio with R2 of 0.99936. The corresponding MSE for this topology was 0.00006. Read More Read Less

Abstract Generally in design of combinational classifier systems, the more diverse the results of the classifiers, the more appropriate final result. In this paper, a new method for combining classifiers is proposed which its main idea is heuristic retraining of classifiers. Specifically, in the new method named Combinational Classifiers using Heuristic Retraining (CCHR) which proposes a new way for generating diversity... Generally in design of combinational classifier systems, the more diverse the results of the classifiers, the more appropriate final result. In this paper, a new method for combining classifiers is proposed which its main idea is heuristic retraining of classifiers. Specifically, in the new method named Combinational Classifiers using Heuristic Retraining (CCHR) which proposes a new way for generating diversity in ensemble pool, a classifier is first run, then, focusing on the drawbacks of this base classifier, other classifiers are retrained heuristically. Each of these classifiers looks at the data with its own attitude. The main concentration in the retrained classifiers is to leverage the error-prone data. So, retrained classifiers usually have different votes about the sample points which are close to boundaries and may be likely erroneous. Experiments show significant improvements in terms of accuracies of consensus classification. This study also investigates that focusing on which crucial data points can lead to more performance in base classifiers. Also, this study shows that adding the number of all “difficult” data points like boosting method, does not always cause a better performance. The experimental results show that the performance of the proposed algorithm outperforms some of the best methods in the literature. So empirically, the authors claim that forcing crucial data points to the training set as well as eliminating them from the training set can yield to the more accurate results, conditionally. Read More Read Less

Abstract Removing stop words is very useful for many text processing applications e.g. text/document retrieval, cross language translation, text categorization, text summarization etc. In this world, different language has different stop word lists, and those are useful for text processing applications. Literature claims that the use of such lists improves retrieval performance. The goal of this research is to evaluate the... Removing stop words is very useful for many text processing applications e.g. text/document retrieval, cross language translation, text categorization, text summarization etc. In this world, different language has different stop word lists, and those are useful for text processing applications. Literature claims that the use of such lists improves retrieval performance. The goal of this research is to evaluate the effect of using English stop word lists in Latent Semantic Indexing (LSI)-based information retrieval (IR) systems with large text dataset. Here, three different lists are compared: two were compiled by IR groups at the University of Glasgow, and the University of Tennessee, and the third one is our own list developed at the University of Northern British Columbia. We also examined the case where stop words were not removed from the input dataset. Our research finds that using tailored stop word lists improves retrieval performance. On the other hand, using arbitrary (non-tailored) lists or not using any list reduces the retrieval performance of LSI-based IR systems with large text dataset. Read More Read Less

Abstract Ensemble methods like Bagging and Boosting which combine the decisions of multiple hypotheses are among the strongest existing machine learning methods. The diversity of the members of an ensemble is known to be an important factor in determining its generalization error. We present a new method for generating ensembles, named CDEBMTE (Creation of Diverse Ensemble Based on Manipulation of Training... Ensemble methods like Bagging and Boosting which combine the decisions of multiple hypotheses are among the strongest existing machine learning methods. The diversity of the members of an ensemble is known to be an important factor in determining its generalization error. We present a new method for generating ensembles, named CDEBMTE (Creation of Diverse Ensemble Based on Manipulation of Training Examples), that directly constructs diverse hypotheses using manipulation of training examples in three ways: (1) sub-sampling training examples, (2) decreasing/increasing errorprone training examples and (3) decreasing/increasing neighbor samples of error-prone training examples. The technique is a simple, general meta-learner that can use any strong learner as a base classifier to build diverse committees. Experimental results using two well-known classifiers (1) decision-tree induction and (2) multilayer perceptron as two base learners demonstrate that this approach consistently achieves higher predictive accuracy than both the base classifier, Adaboost and Bagging. CDEBMTE also outperforms Adaboost more prominent when training data size is becomes larger. We propose to show that CDEBMTE can be effectively used to achieve higher accuracy and to obtain better class membership probability estimates. Read More Read Less

Abstract Increasingly web is used for tracking numerous information pieces of large and complex data. Normal way of accessing these sources using web search engines has many limitations and we have enough literature to document them. This study is an enhancement to trap the discovery of targeted information pieces which are normally failed cases in search engines. We use an integrated... Increasingly web is used for tracking numerous information pieces of large and complex data. Normal way of accessing these sources using web search engines has many limitations and we have enough literature to document them. This study is an enhancement to trap the discovery of targeted information pieces which are normally failed cases in search engines. We use an integrated approach of conventional web search and semantic web by carefully constructed algorithms. The algorithms are tested to track data about the gulf institutions. The experiments provide encouraging results and enable us to draw future directions of research in this domain. Read More Read Less