We presented a comparison between several feature ranking methods used on two real datasets. We considered six ranking methods that can be divided into two broad categories: statistical and entropy-based. Four supervised learning algorithms are adopted to build models, namely, IB1, Naive Bayes, C4.5 decision tree and the RBF network. We showed that the selection of ranking methods could be important for classification accuracy. In our experiments, ranking methods with different supervised learning algorithms give quite different results for balanced accuracy. Our cases confirm that, in order to be sure that a subset of features giving the highest accuracy has been selected, the use of many different indices is recommended
This paper deals with the comparative analysis of prediction classifiers in the blended learning environment. The model proposed in this paper predicts students’ final grades based on activities within different educational environments. A comparative study of classifier performance has been performed in order to determine the classifier most suitable for multiclass feature dataset. Important results for different classes have been obtained using different classifiers, and the majority vote scheme is subsequentially used to form an ensemble based on Naïve Bayes, Hidden Naïve Bayes, J48 decision tree and Random Forest. According to experimental evaluation, there is a significant improvement of proposed model's precision and accuracy regarding the students’ grades prediction in blended learning environment scenario. The major contribution of the research presented in this paper is an efficient multi‐class prediction model applicable to aforementioned environment.
Although the importance of contextual information in speech recognition has been acknowledged for a long time now, it has remained clearly underutilized even in state-of-the-art speech recognition systems. This article introduces a novel, methodologically hybrid approach to the research question of context-dependent speech recognition in human-machine interaction. To the extent that it is hybrid, the approach integrates aspects of both statistical and representational paradigms. We extend the standard statistical pattern-matching approach with a cognitively inspired and analytically tractable model with explanatory power. This methodological extension allows for accounting for contextual information which is otherwise unavailable in speech recognition systems, and using it to improve post-processing of recognition hypotheses. The article introduces an algorithm for evaluation of recognition hypotheses, illustrates it for concrete interaction domains, and discusses its implementation within two prototype conversational agents.
Upgraded Petri nets and a two-level edge detection algorithm for the modeling, simulation, detection, and analysis of textile pilling has used in this paper. There are two models in this paper. The first model is related to creating images with pilling on textiles. This model is represented at a low level of abstraction, using parallelism, synchronization, and preservation of model markup. After several iterations of the modeling, simulation, and analysis, the model transformed into a specific Python program. This program generates several hundred images of pilling on textiles. The second model refers to the modeling, simulation, and analysis of the preparation of generated images for entry into a program that uses a two-level algorithm for edge detection and generation of output images of detected pilling. An image preparation includes comparing pixel blocks (2x2, 4x4, and 8x8) of the image with textile pilling and the image without pilling. This image with pilling is modified regardless of the difference between the corresponding blocks. This image is an input to the two-level edge detection algorithm. The paper presents some examples of generated images of textile pilling and images with detected pilling. Errors due to the size of the comparison block about the exact pilling surface shown. The speed of image processing according to the size of the comparison blocks analyzed. Some guidelines represented for further research on pilling detection using Upgraded Petri nets and a two-level edge detection algorithm.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.