Search

The aim of the study is to investigate the possibility of applying machine learning methods for saccades and fixation detection from eye tracking signals. There are two widely used approaches to solving this task: velocity-based saccade identification (I-VT) and dispersion-based saccade identification (I-DT). Both require setting an optimal threshold which directly influences detection quality. Another important problem comes from the necessity of noise handling if eye tracking signal has a high noise level. Using I-VT and I-DT algorithms in this case will result in misclassification. Previous researchers have obtained promising results using Random Forest algorithm for saccade and fixation detection. Our aim with the article was to compare Random Forest with other binary classification algorithms such as Logistic Regression and K-Nearest Neighbors. We've also compared mentioned algorithms with velocity-based approach. The results show that Random Forest has the best classification metrics. The findings suggest that this machine learning approach could also be useful for eye movement event detection and proves the importance of further research in this field [ABSTRACT FROM AUTHOR]

Sentiment, emotions, feelings, showing or giving judgments by face gestures, etc. can be communicated by text, speech or images. Analyzing image Polarity via images is now a burgeoning research field. An image can easily be interpreted by any human being and acts as a bridge to relate an image to the human's thoughts. The paper proposed a model for analyzing the sentiments or emotions of an image. We have used a Data-Mining Orange Tool for executing the experiment for classification and sentiment analysis of an image. The main challenge is to acquire the polarity of the unlabeled data using optimal machine learning techniques. To overcome this challenge for getting the polarity and sentiment of an unlabeled data the proposed model used various noteworthy machine learning models like Neural Network, Naïve Bayes and random forest for classifying and analyzing the image sentiment and also provide a comparison based on certain parameters like Area Under Curve, Precision, Recall, classification Accuracy, etc. The tool provides a complete package for performing image analysis using the widgets available with orange tool and a data visualization tool to get a visible representation of the output. The proposed model shows that Random Forest performance the best from the three compared algorithms with classification Accuracy of 0.747, recall & precision as 0.747 & 0.752 and the accuracy based on confusion matrix is 74.72%. [ABSTRACT FROM AUTHOR]

Imbalance of dataset in the accuracy testing process can lead to biased results. It occurs due to insufficient data in the training phase where unbalanced data causes problems in Machine Learning. Classification and predicting results become difficult when there is insufficient data to study. To overcome this, it takes steps in balancing the data, one of which is the random over sampling technique. The basic principle of using this technique is to rebalance an unbalanced data set with a concrete strategy. The use of sampling technique in the case of data imbalance is proven to be able to improve the performance of the algorithm. The results of testing the KNN, Naive Bayes, SVM, J.48 and Random Forest algorithms using 10 fold cross validation on the public dataset of early stage diabetes risk prediction from the Hospital in Sylhet, Bangladesh after going through the re-sample stage proved to be able to improve measurement results with a high level of accuracy. The highest using KNN with an accuracy of 99.4231%, while the results of the J.48 algorithm test with an accuracy rate of 99.2308%, Random Forest with an accuracy rate of 98.0769%, SVM with an accuracy rate of 96.5385% and Naïve Bayes with an accuracy rate of 90.3846%. [ABSTRACT FROM AUTHOR]

The process of building new dataset and the existence of such a data followed the urgent need for the existence of datasets that are specialized in educational lectures, so this will need an accurate classification algorithm to classify it, the benefit of classify such dataset is to minimize the workload of classifying each file manually and individually. In the present paper, authors perform experimentations for conducting an empirical deep learning study, especially, convolutional neural network, for three new datasets of educational lectures which are (PDF, Word and PowerPoint datasets), The three new datasets using real data educational resources lectures collected from various document projects of different universities and institutions. The architecture has been applied to the task of the text classification in the domain of the document with documents data-sets have been obtained from a variety of projects on actual document cases. The aim of the present study is to initially test the performance of each dataset (PDF, Word, and PowerPoint dataset) through using four machine learning classification algorithms which are (Bayes Net, Random Forest, Random Committee, and OneR). Second goal is experimenting the efficiency of the approach of the deep learning in the tasks of classification and after that, comparing the efficiencies with the efficiencies of traditional machine learning classification methods. Mainly two classification techniques used to maximize the benefits of the classification process, the first one is to use the deep learning algorithm which shows an accuracy of classifying file between (95 and 96%) for three new dataset files and standard machine learning algorithms (OneR, Random forest, Bayes net, and Random Committee) these algorithm shows accuracy 91% for PDF Dataset using random forest and random committee algorithms, for Word dataset the accuracy is 46% using random committee, and for the last dataset PowerPoint the accuracy is 77% using random forest, Therefore, we will choose Deep learning algorithm because it gives higher results and accuracy than machine learning algorithms. [ABSTRACT FROM AUTHOR]

Islamophobia is explored in many different perspectives in various settings, however Internet-based or online settings Islamophobia remain under-researched. Islamophobic content on social media promotes a biased and negative sentiments towards Muslims. The aim of this paper is to develop an automatic recognition and detection of Islamophobic tweets to address bigotry and hatred towards Islam and Muslims. A unique set of lexical, syntactic and semantic features were extracted from 2200 tweets containing keywords "Islam" and "Muslims". Using Sequential Minimal Optimisation (SMO), and Logistic Regression (LR), we achieved f-measures of above 0.7, demonstrating the accuracy of the classifier and the usefulness of the features extracted. Based on the experiments conducted, Random Forest (RF) algorithm proved to be the top performing algorithm in this study showing 86.69% of f-measure in training model, with LibSVM being the second top performing algorithm. [ABSTRACT FROM AUTHOR]

The implementation of high-speed broadband has resulted in a huge rise in internet use, which has skyrocketed as more people use remote access. Despite the fact that this rise in use allows for more online activities, the rapid growth of the Internet provides attackers with a wealth of network knowledge, and a large number of network threats has a negative effect on the network's development. Machine learning approaches are good at detecting unknown attacks. Random Forest, Decision Tree, Support Vector Machine, Logistic Regression, Gaussian Naive Bayes, and Gradient boosting Classifier are among the Supervised Machine Learning classification algorithms evaluated against the NSL-KDD dataset in this paper. Random Forest Classifier outperforms other methods in detecting attacks, according to the results of the experiments. [ABSTRACT FROM AUTHOR]

With the continuous development of machine learning, industry and academia have released a lot of machine learning frameworks based on distributed computing platform, and have been widely used. However, the existing framework of machine learning is limited by the limitations of machine learning algorithm itself, such as the choice of parameters and the interference of noises, the high using threshold and so on. This paper introduces the research background of machine learning framework, and combined with the commonly used random forest algorithm in machine learning classification algorithm, puts forward the research objectives and content, proposes an improved adaptive random forest algorithm (referred to as ARF), and on the basis of ARF, designs and implements the machine learning framework. [ABSTRACT FROM AUTHOR]

This paper presents an image classification algorithm based on spatial sparse coding model and random forest. Firstly, SIFT feature extraction of the image; and then use the sparse encoding theory to generate visual vocabulary based on SIFT features, and using the visual vocabulary of SIFT features into a sparse vector; through the combination of regional integration and spatial sparse vector, the sparse vector gets a fixed dimension is used to represent the image; at last random forest classifier for image sparse vectors for training and testing, using the experimental data set for standard test Caltech-101 and Scene-15. The experimental results show that the proposed algorithm can effectively represent the features of the image and improve the classification accuracy. In this paper, we propose an innovative image recognition algorithm based on image segmentation, sparse coding and multi instance learning. This algorithm introduces the concept of multi instance learning, the image as a multi instance bag, sparse feature transformation by SIFT images as instances, sparse encoding model generation visual vocabulary as the feature space is mapped to the feature space through the statistics on the number of instances in bags, and then use the 1-norm SVM to classify images and generate sample weights to select important image features. [ABSTRACT FROM AUTHOR]