IJMECS Vol. 13, No. 3, Jun. 2021
Cover page and Table of Contents: PDF (size: 598KB)
This study was conducted to evaluate the effects of a parenting training program that supports language acquisition in early childhood. To reveal such an effect, an experimental research design including pre-test-post-test and retention test was applied respectively. Experimental and control groups were formed with the parents of 4-6 years old children attending pre-school education institutions. To assess the language development levels of the children, Peabody Picture-Vocabulary Test (PPVT) was applied during the pre-test phase; after the parental training, a post-test was applied; and a year later, a retention test was implemented alternately. Parents in the experimental group evaluated the program after the Parental Support Program (PSP). The personal characteristics of the study group and the opinions of the parents evaluating the training have been shown by using the frequency and percentages. Whether PPVT and PSP scores differ according to socio-demographic variables was analyzed by t-tests. In the end, there was a significant increase in the results of the post-test and retention test performed after parent training that supports language acquisition. This increase has been found to be significantly higher than the PPVT scores of children in the control group. Thus, we have determined that the parents have a positive attitude towards the training program. The results of the study also reveal that parenting training that supports children's language acquisition has a positive effect on children's language development.[...] Read more.
One of the trends in information technologies is implementing neural networks in modern software packages . The fact that neural networks cannot be directly programmed (but trained) is their distinctive feature. In this regard, the urgent task is to ensure sufficient speed and quality of neural network training procedures. The process of neural network training can differ significantly depending on the problem. There are verification methods that correspond to the task’s constraints; they are used to assess the training results. Verification methods provide an estimate of the entire cardinal set of examples but do not allow to estimate which subset of those causes a significant error. This fact leads to neural networks’ failure to perform with the given set of hyperparameters, making training a new one time-consuming.
On the other hand, existing empirical assessment methods of neural networks training use discrete sets of examples. With this approach, it is impossible to say that the network is suitable for classification on the whole cardinal set of examples.
This paper proposes a criterion for assessing the quality of classification results. The criterion is formed by describing the training states of the neural network. Each state is specified by the correspondence of the set of errors to the function range representing a cardinal set of test examples. The criterion usage allows tracking the network’s classification defects and marking them as safe or unsafe. As a result, it is possible to formally assess how the training and validation data sets must be altered to improve the network’s performance, while existing verification methods do not provide any information on which part of the dataset causes the network to underperform.
Indonesia has been known as an agrarian country because of its fertile soil and is very suitable for agricultural land, including rice. Yogyakarta is one of the most significant granary regions in Indonesia, especially in the Sleman region. However, one of the main challenges in rice planting in recent years is the erratic rainfall patterns caused by climate anomalies due to the El Nino and La Nina phenomena. As a result of this phenomenon, farmers have difficulty determining planting time and harvest time and planting other plants. Therefore, we make rainfall predictions to recommend planting varieties with Moving Average and Naive Bayes Methods in Sleman District. The results showed that moving averages well use in predicting rainfall. From these results, we can estimate that in 2020 rice production will below. That can saw from the calculation of the probability of naive Bayes on rice plants being low at 0.999 and 0.923. So that the recommended intercrops planted in 2020 are corn and peanuts. We also find that rainfall prediction with Moving Average using data from several previous years in the same month is more accurate than using data from four past months or periods.[...] Read more.
Code clone detection plays a vital role in both industry and academia. Last three decades have seen more than 250 clone detection techniques with lack of single framework that can detect and classify all 4 basic types of code clones with high precision. This serious lack of clone classification impacts largely on the universities and online learning platforms that fail to validate the projects or coding assignments submitted online. In this paper, we propose a complete and language agnostic technique to detect and classify all 4 clone types of C, C++, and Java programs. The method first generates the parse tree then extracts the functional tree to eliminate the need for the preprocessing stage employed by previous clone detection techniques. The generated parse tree contains all the necessary information for detecting code clones. We employ TF-IDF cosine similarity for the proper classification of clone types. The proposed technique achieves incredible precision rate of 100% in detecting the first two types of clones and 98% precision in detecting type-3 and type-4 clones for small codes of C, C++, and Java containing an average line count of 5. The proposed technique outperforms the existing tree-based clone detection tools by providing the average precision of 98.07% on the C, C++, and Java programs crawled from Github with an average line count of 15 which signifies that cosine similarity measure on ANTLR functional tree accurately detects all 4 types of small clones and act as proper validation tools for identifying the learning level in the submitted programming assignment.[...] Read more.
Data mining approaches provide different educational institutions opportunities to find hidden patterns from the data stored in the database. Many researchers have used these data to develop a model that would assist the institution administrators in decision-making. This study was performed to predict student program completion using the Naïve Bayes classifier technique. The dataset utilized in this study was obtained from Bulacan State University – Sarmiento Campus in the Philippines under BS Information Technology program from five-year graduates’ data for Academic Year 2012-2016. This dataset was pre-processed, cleansed, transformed, and balanced before constructing the model. Ten predictors were used for predicting student completion. The feature selection technique was used to filter and evaluate the significance of each factor. The significant variables assessed by the feature selection technique (Weight by Correlation) were the final parameters in creating the model. The Naïve Bayes classifier was applied to predict the students’ completion using the 70:30 ratios for training and testing dataset distribution. Correlation analysis identified the weight of individual attributes to the label attribute. From 10 possible predictor variables, only four (4) predictor variables were selected after correlation analysis. The identified significant attributes affecting program completion, namely (in order of significance): parents' monthly income, mother and father's educational attainment, and High School GPA attributes. The significant attributes identified in correlation analysis splitted into 70% training data or 447 records and 30% testing data or 191 records. There were 84 out of 191 data samples, or 44% of students were predicted to complete the program. On the other hand, 107 out of 191 data samples, or 56%, were predicted as not completing the program. The accuracy values performed an 84% rating with 80.46% class precision, and 83.33% class recall in the testing dataset (n=191). The outcomes of this study have a significant impact on HEIs, particularly on college completion rates. This study shall be highly significant and beneficial specifically to university administrators as this be a tool for them to identify students who will complete college based on variables included in the model.[...] Read more.
In this study the authors investigated the connections between the training processes of unsupervised neural network models with self-encoding and regeneration and the information structure in the representations created by such models. We propose theoretical arguments leading to conclusions, confirmed by previously published experimental results that unsupervised representations obtained under certain constraints in training compliant with Bayesian inference principle, favor configurations with better categorization of hidden concepts in the observable data. The results provide an important connection between training of unsupervised machine learning models and the structure of representations created by them and can be used in developing new methods and approaches in self-learning as well as provide insights into common principles underlying the emergence of intelligence in machine and biologic systems.[...] Read more.