Computer Engineering / Bilgisayar Mühendisliği
Permanent URI for this collectionhttps://hdl.handle.net/11147/10
Browse
6 results
Search Results
Article Performance and Accuracy Predictions of Approximation Methods for Shortest-Path Algorithms on Gpus(Elsevier, 2022) Aktılav, Busenur; Öz, IşılApproximate computing techniques, where less-than-perfect solutions are acceptable, present performance-accuracy trade-offs by performing inexact computations. Moreover, heterogeneous architectures, a combination of miscellaneous compute units, offer high performance as well as energy efficiency. Graph algorithms utilize the parallel computation units of heterogeneous GPU architectures as well as performance improvements offered by approximation methods. Since different approximations yield different speedup and accuracy loss for the target execution, it becomes impractical to test all methods with various parameters. In this work, we perform approximate computations for the three shortest-path graph algorithms and propose a machine learning framework to predict the impact of the approximations on program performance and output accuracy. We evaluate random predictions for both synthetic and real road-network graphs, and predictions of the large graph cases from small graph instances. We achieve less than 5% prediction error rates for speedup and inaccuracy values.Article Citation - WoS: 4Citation - Scopus: 4Predicting the Soft Error Vulnerability of Parallel Applications Using Machine Learning(Springer, 2021) Öz, Işıl; Arslan, SanemWith the widespread use of the multicore systems having smaller transistor sizes, soft errors become an important issue for parallel program execution. Fault injection is a prevalent method to quantify the soft error rates of the applications. However, it is very time consuming to perform detailed fault injection experiments. Therefore, prediction-based techniques have been proposed to evaluate the soft error vulnerability in a faster way. In this work, we present a soft error vulnerability prediction approach for parallel applications using machine learning algorithms. We define a set of features including thread communication, data sharing, parallel programming, and performance characteristics; and train our models based on three ML algorithms. This study uses the parallel programming features, as well as the combination of all features for the first time in vulnerability prediction of parallel programs. We propose two models for the soft error vulnerability prediction: (1) A regression model with rigorous feature selection analysis that estimates correct execution rates, (2) A novel classification model that predicts the vulnerability level of the target programs. We get maximum prediction accuracy rate of 73.2% for the regression-based model, and achieve 89% F-score for our classification model.Book Part Citation - WoS: 299Citation - Scopus: 406Introduction To Machine Learning(Humana Press, 2014) Baştanlar, Yalın; Özuysal, MustafaThe machine learning field, which can be briefly defined as enabling computers make successful predictions using past experiences, has exhibited an impressive development recently with the help of the rapid increase in the storage capacity and processing power of computers. Together with many other disciplines, machine learning methods have been widely employed in bioinformatics. The difficulties and cost of biological analyses have led to the development of sophisticated machine learning approaches for this application area. In this chapter, we first review the fundamental concepts of machine learning such as feature assessment, unsupervised versus supervised learning and types of classification. Then, we point out the main issues of designing machine learning experiments and their performance evaluation. Finally, we introduce some supervised learning methods.Conference Object Citation - WoS: 18Citation - Scopus: 25Adaptation and Use of Subjectivity Lexicons for Domain Dependent Sentiment Classification(Institute of Electrical and Electronics Engineers Inc., 2012) Dehkharghani, Rahim; Yanıkoğlu, Berrin; Tapucu, Dilek; Saygın, YücelSentiment analysis refers to the automatic extraction of sentiments from a natural language text. We study the effect of subjectivity-based features on sentiment classification on two lexicons and also propose new subjectivity-based features for sentiment classification. The subjectivity-based features we experiment with are based on the average word polarity and the new features that we propose are based on the occurrence of subjective words in review texts. Experimental results on hotel and movie reviews show an overall accuracy of about 84% and 71% in hotel and movie review domains respectively; improving the baseline using just the average word polarities by about 2% points. © 2012 IEEE.Conference Object Citation - WoS: 22Citation - Scopus: 29Learning Domain-Specific Polarity Lexicons(Institute of Electrical and Electronics Engineers Inc., 2012) Demiröz, Gülşen; Yanıkoğlu, Berrin; Tapucu, Dilek; Saygın, YücelSentiment analysis aims to automatically estimate the sentiment in a given text as positive or negative. Polarity lexicons, often used in sentiment analysis, indicate how positive or negative each term in the lexicon is. However, since creating domain-specific polarity lexicons is expensive and time-consuming, researchers often use a general purpose or domain-independent lexicon. In this work, we address the problem of adapting a general purpose polarity lexicon to a specific domain and propose a simple yet effective adaptation algorithm. We experimented with two sets of reviews from the hotel and movie domains and observed that while our adaptation techniques changed the polarity values for only a small set of words, the overall test accuracy increased significantly: 77% to 83% in the hotel dataset and 61% to 66% in the movie dataset. © 2012 IEEE.Conference Object Citation - WoS: 2Citation - Scopus: 5Machine Learning Based Learner Modeling for Adaptive Web-Based Learning(Springer Verlag, 2007) Aslan, Burak Galip; İnceoğlu, Mustafa MuratEspecially in the first decade of this century, learner adapted interaction and learner modeling are becoming more important in the area of web-based learning systems. The complicated nature of the problem is a serious challenge with vast amount of data available about the learners. Machine learning approaches have been used effectively in both user modeling, and learner modeling implementations. Recent studies on the challenges and solutions about learner modeling are explained in this paper with the proposal of a learner modeling framework to be used in a web-based learning system. The proposed system adopts a hybrid approach combining three machine learning techniques in three stages.