Tomche Delev - Academia.edu (original) (raw)

Papers by Tomche Delev

Lecture Notes in Computer Science, 2015

Motivated by an increasing number of new applications, the research community is devoting an incr... more Motivated by an increasing number of new applications, the research community is devoting an increasing amount of attention to the task of multi-label classification (MLC). Many different approaches to solving multi-label classification problems have been recently developed. Recent empirical studies have comprehensively evaluated many of these approaches on many datasets using different evaluation measures. The studies have indicated that the predictive performance and efficiency of the approaches could be improved by using data derived (artificial) hierarchies, in the learning and prediction phases. In this paper, we compare different clustering algorithms for constructing the label hierarchies (in a data-driven manner), in multi-label classification. We consider flat label sets and construct the label hierarchies from the label sets that appear in the annotations of the training data by using four different clustering algorithms (balanced k-means, agglomerative clustering with single and complete linkage and predictive clustering trees). The hierarchies are then used in conjunction with global hierarchical multi-label classification (HMC) approaches. The results from the statistical and experimental evaluation reveal that the data-derived label hierarchies used in conjunction with global HMC methods greatly improve the performance of MLC methods. Additionally, multi-branch hierarchies appear much more suitable for the global HMC approaches as compared to the binary hierarchies.

Implementing a web-based system for automatic assessment is a big step in the introductionary pro... more Implementing a web-based system for automatic assessment is a big step in the introductionary programming courses. In this paper we study and report the data generated by the usage of the system Code developed at the Faculty of Computer Science and Engineering. The system supports compilation and execution of programming problems in exercises and exams and it is used in many courses that involve programming assignments. The analyzed data shows the differences in working in laboratory settings, compared to practical exams. We also present the results from plagiarism detection, and report significant cases of plagiarism in introductionary courses. At the end we present the results from initial qualitative evaluation of the system by surveying 48 students.

E-Lab is a system developed at Faculty of Computer Science and Engineering for solving and auto-g... more E-Lab is a system developed at Faculty of Computer Science and Engineering for solving and auto-grading programming problems from introduction to programming courses. The main goal is to simplify and improve the organization and the process of solving programming problems from large group of students in dedicated computer labs using centralized server. All the work from the students is done in a web browser using a web-based code editor and everything is stored, compiled and executed on the server. The system keeps records of all problem attempts from identified students which are used as attendance records. All the problems and solutions are under version control system (Git). The platform supports different types of problems in several programming languages (C, C++, Java) and it's designed to be easily extended.

ICT Innovations 2010, 2010

Abstract. Ensemble methods are able to improve the predictive performance of many base classifier... more Abstract. Ensemble methods are able to improve the predictive performance of many base classifiers. In this paper, we consider two ensemble learning techniques, bagging and random forests, and apply them to Binary SVM Decision Tree (SVM-BDT). Binary SVM Decision Tree is a tree based architecture that utilizes support vector machines for solving multiclass problems. It takes advantage of both the efficient computation of the decision tree architecture and the high classification accuracy of SVMs. In this paper we empirically ...

This paper presents the design and implementation of a mobile application along with a web server... more This paper presents the design and implementation of a mobile application along with a web server for geo-tagging favorite and interesting places and sharing them with the community. The design and architecture shows some key aspects and issues concerning this kind of system. The mobile application is implemented in J2ME and tested on GPS enabled Nokia phones and the web server is implemented on cloud infrastructure implementation, the Google App Engine. The system was evaluated with real devices and ...

This article presents an overview about the basis of estimators of connectivity, such as Directed... more This article presents an overview about the basis of estimators of connectivity, such as Directed Transfer Function (DTF) and Partial Directed Coherence (PDC), their differences and their applicability in the estimation and analysis of functional brain connectivity.

AI 2010: Advances in Artificial Intelligence: 23rd Australasian Joint Conference, Adelaide, Australia, December 7-10, 2010. Proceedings, Dec 14, 2010

Lecture Notes in Computer Science, 2015

Motivated by an increasing number of new applications, the research community is devoting an incr... more Motivated by an increasing number of new applications, the research community is devoting an increasing amount of attention to the task of multi-label classification (MLC). Many different approaches to solving multi-label classification problems have been recently developed. Recent empirical studies have comprehensively evaluated many of these approaches on many datasets using different evaluation measures. The studies have indicated that the predictive performance and efficiency of the approaches could be improved by using data derived (artificial) hierarchies, in the learning and prediction phases. In this paper, we compare different clustering algorithms for constructing the label hierarchies (in a data-driven manner), in multi-label classification. We consider flat label sets and construct the label hierarchies from the label sets that appear in the annotations of the training data by using four different clustering algorithms (balanced k-means, agglomerative clustering with single and complete linkage and predictive clustering trees). The hierarchies are then used in conjunction with global hierarchical multi-label classification (HMC) approaches. The results from the statistical and experimental evaluation reveal that the data-derived label hierarchies used in conjunction with global HMC methods greatly improve the performance of MLC methods. Additionally, multi-branch hierarchies appear much more suitable for the global HMC approaches as compared to the binary hierarchies.

Implementing a web-based system for automatic assessment is a big step in the introductionary pro... more Implementing a web-based system for automatic assessment is a big step in the introductionary programming courses. In this paper we study and report the data generated by the usage of the system Code developed at the Faculty of Computer Science and Engineering. The system supports compilation and execution of programming problems in exercises and exams and it is used in many courses that involve programming assignments. The analyzed data shows the differences in working in laboratory settings, compared to practical exams. We also present the results from plagiarism detection, and report significant cases of plagiarism in introductionary courses. At the end we present the results from initial qualitative evaluation of the system by surveying 48 students.

E-Lab is a system developed at Faculty of Computer Science and Engineering for solving and auto-g... more E-Lab is a system developed at Faculty of Computer Science and Engineering for solving and auto-grading programming problems from introduction to programming courses. The main goal is to simplify and improve the organization and the process of solving programming problems from large group of students in dedicated computer labs using centralized server. All the work from the students is done in a web browser using a web-based code editor and everything is stored, compiled and executed on the server. The system keeps records of all problem attempts from identified students which are used as attendance records. All the problems and solutions are under version control system (Git). The platform supports different types of problems in several programming languages (C, C++, Java) and it's designed to be easily extended.

ICT Innovations 2010, 2010

Abstract. Ensemble methods are able to improve the predictive performance of many base classifier... more Abstract. Ensemble methods are able to improve the predictive performance of many base classifiers. In this paper, we consider two ensemble learning techniques, bagging and random forests, and apply them to Binary SVM Decision Tree (SVM-BDT). Binary SVM Decision Tree is a tree based architecture that utilizes support vector machines for solving multiclass problems. It takes advantage of both the efficient computation of the decision tree architecture and the high classification accuracy of SVMs. In this paper we empirically ...

This paper presents the design and implementation of a mobile application along with a web server... more This paper presents the design and implementation of a mobile application along with a web server for geo-tagging favorite and interesting places and sharing them with the community. The design and architecture shows some key aspects and issues concerning this kind of system. The mobile application is implemented in J2ME and tested on GPS enabled Nokia phones and the web server is implemented on cloud infrastructure implementation, the Google App Engine. The system was evaluated with real devices and ...

This article presents an overview about the basis of estimators of connectivity, such as Directed... more This article presents an overview about the basis of estimators of connectivity, such as Directed Transfer Function (DTF) and Partial Directed Coherence (PDC), their differences and their applicability in the estimation and analysis of functional brain connectivity.

AI 2010: Advances in Artificial Intelligence: 23rd Australasian Joint Conference, Adelaide, Australia, December 7-10, 2010. Proceedings, Dec 14, 2010