scispace - formally typeset
Search or ask a question
Author

Erind Bedalli

Bio: Erind Bedalli is an academic researcher from Aleksandër Xhuvani University. The author has contributed to research in topics: Fuzzy clustering & Fuzzy logic. The author has an hindex of 3, co-authored 7 publications receiving 24 citations. Previous affiliations of Erind Bedalli include University of Tirana & Epoka University.

Papers
More filters
Journal ArticleDOI
TL;DR: This paper is providing a heterogeneous cluster ensemble approach to improve the stability of fuzzy cluster analysis by applying different fuzzy clustering algorithms on the datasets obtaining multiple partitions, which in the later stage will be fused into the final consensus matrix.

10 citations

Book ChapterDOI
12 Sep 2012
TL;DR: Experiments over long time series datasets from the UCR collection demonstrate the superiority of the proposed fast linear classification complexity method, which is orders of magnitude faster than baselines while being superior even in terms of classification accuracy.
Abstract: Time-series classification has gained wide attention within the Machine Learning community, due to its large range of applicability varying from medical diagnosis, financial markets, up to shape and trajectory classification. The current state-of-art methods applied in time-series classification rely on detecting similar instances through neighboring algorithms. Dynamic Time Warping (DTW) is a similarity measure that can identify the similarity of two time-series, through the computation of the optimal warping alignment of time point pairs, therefore DTW is immune towards patterns shifted in time or distorted in size/shape. Unfortunately the classification time complexity of computing the DTW distance of two series is quadratic, subsequently DTW based nearest neighbor classification deteriorates to quartic order of time complexity per test set. The high time complexity order causes the classification of long time series to be practically infeasible. In this study we propose a fast linear classification complexity method. Our method projects the original data to a reduced latent dimensionality using matrix factorization, while the factorization is learned efficiently via stochastic gradient descent with fast convergence rates and early stopping. The latent data dimensionality is set to be as low as the cardinality of the label variable. Finally, Support Vector Machines with polynomial kernels are applied to classify the reduced dimensionality data. Experimentations over long time series datasets from the UCR collection demonstrate the superiority of our method, which is orders of magnitude faster than baselines while being superior even in terms of classification accuracy.

8 citations

Journal ArticleDOI
TL;DR: This study has found the application of several variations of the fuzzy clustering algorithms on these data to be a productive approach and will utilize several other variations like the possibilistic fuzzy c-means, the Gustafson-Kessel algorithm and the kernel based fuzzy clusters.
Abstract: Clustering is a very useful technique which helps to enrich the semantics of the data by revealing patterns in large collections of poly-dimensional data. Moreover the fuzzy approach in clustering provides flexibility and enhanced modeling capability, as the results are expressed in soft clusters, allowing partial memberships of data points in the clusters. During the last decade, the digitalization of detailed student records of the University of Elbasan has not only simplified the typical university procedures but also it has created the possibility of a deeper view of the students’ data. The cluster analysis applied on these student data can discover patterns which would assist in several strategic issues like: optimizing the student advising process, organization of curricula, adjusting the compulsory/elective courses, preparing better teaching approaches etc. In our study, besides the classical fuzzy c-means, we will utilize several other variations like the possibilistic fuzzy c-means, the Gustafson-Kessel algorithm and the kernel based fuzzy clustering. We have found the application of several variations of the fuzzy clustering algorithms on these data to be a productive approach. Particular applications sometimes provide useful viewpoints which trigger innovative ideas for the policy-makers of the university.

5 citations

Book ChapterDOI
05 Sep 2018
TL;DR: A hybrid fuzzy clustering model combining variants of fuzzy c-means clustering and density based clustering for exploring well-structured user feedback data intending to exploit the advantages of these two types of clustering approaches and diminishing their drawbacks is presented.
Abstract: In today’s dynamic environments, user feedback data are a valuable asset providing orientations about the achieved quality and possible improvements of various products or services. In this paper we will present a hybrid fuzzy clustering model combining variants of fuzzy c-means clustering and density based clustering for exploring well-structured user feedback data. Despite of the multitude of successful applications where these algorithms are applied separately, they also suffer drawbacks of various kinds. So, the FCM algorithm faces difficulties in detecting clusters of non-spherical shapes or densities and moreover it is sensitive to noise and outliers. On the other hand density-based clustering is not easily adaptable to generate fuzzy partitions. Our hybrid clustering model intertwines density-based clustering and variations of FCM intending to exploit the advantages of these two types of clustering approaches and diminishing their drawbacks. Finally we have assessed and compared our model in a real-world case study.

3 citations

Book ChapterDOI
13 May 2014
TL;DR: This paper innovatively extends existing factorization models into a supervised nonlinear factorization, and jointly reconstructs both the observed predictors and target variables via generative-style nonlinear functions.
Abstract: Semi-supervised learning is an eminent domain of machine learning focusing on real-life problems where the labeled data instances are scarce. This paper innovatively extends existing factorization models into a supervised nonlinear factorization. The current state of the art methods for semi-supervised regression are based on supervised manifold regularization. In contrast, the latent data constructed by the proposed method jointly reconstructs both the observed predictors and target variables via generative-style nonlinear functions. Dual-form solutions of the nonlinear functions and a stochastic gradient descent technique which learns the low dimensionality data are introduced. The validity of our method is demonstrated in a series of experiments against five state-of-art baselines, clearly improving the prediction accuracy in eleven real-life data sets.

2 citations


Cited by
More filters
Book
01 Jan 2008
TL;DR: This paper presents a meta-modelling framework for local Multivariate Analysis based on Fuzzy Clustering and Probabilistic PCA Model that automates the very labor-intensive and therefore time-heavy and expensive process of fuzzy clustering.
Abstract: BasicMethods for c-Means Clustering- Variations and Generalizations - I- Variations and Generalizations - II- Miscellanea- Application to Classifier Design- Fuzzy Clustering and Probabilistic PCA Model- Local Multivariate Analysis Based on Fuzzy Clustering- Extended Algorithms for Local Multivariate Analysis

52 citations

Journal ArticleDOI
TL;DR: A novel fuzzy clustering ensemble framework based on a new fuzzy diversity measure and a fuzzy quality measure to find the base-clusterings with the best performance and the effectiveness of the proposed approach compared to the state-of-the-art methods in terms of evaluation criteria on various standard datasets is revealed.
Abstract: In spite of some attempts at improving the quality of the clustering ensemble methods, it seems that little research has been devoted to the selection procedure within the fuzzy clustering ensemble. In addition, quality and local diversity of base-clusterings are two important factors in the selection of base-clusterings. Very few of the studies have considered these two factors together for selecting the best fuzzy base-clusterings in the ensemble. We propose a novel fuzzy clustering ensemble framework based on a new fuzzy diversity measure and a fuzzy quality measure to find the base-clusterings with the best performance. Diversity and quality are defined based on the fuzzy normalized mutual information between fuzzy base-clusterings. In our framework, the final clustering of selected base-clusterings is obtained by two types of consensus functions: (1) a fuzzy co-association matrix is constructed from the selected base-clusterings and then, a single traditional clustering such as hierarchical agglomerative clustering is applied as consensus function over the matrix to construct the final clustering. (2) a new graph based fuzzy consensus function. The time complexity of the proposed consensus function is linear in terms of the number of data-objects. Experimental results reveal the effectiveness of the proposed approach compared to the state-of-the-art methods in terms of evaluation criteria on various standard datasets.

46 citations

Journal ArticleDOI
TL;DR: Experimental results on various standard datasets demonstrated the effectiveness of the proposed approach compared to the state-of-the-art methods in terms of evaluation criteria and clustering robustness.

41 citations

Journal ArticleDOI
TL;DR: A new parameter-free measure for the specific purpose of quickly and accurately assessing the similarity between two given long time series, which outperforms DTW while providing competitive results against popular distance-based classifiers and is orders of magnitude faster than DTW.
Abstract: The problem of similarity measures is a major area of interest within the field of time series classification (TSC). With the ubiquitous of long time series and the increasing demand for analyzing them on limited resource devices, there is a crucial need for efficient and accurate measures to deal with such kind of data. In fact, there are a plethora of good time series similarity measures in the literature. However, most existing methods achieve good performance for short time series, but their effectiveness decreases quickly as time series are longer. In this paper, we develop a new parameter-free measure for the specific purpose of quickly and accurately assessing the similarity between two given long time series. The proposed “Local Extrema Dynamic Time Warping” (LE-DTW) consists of two steps. The first is a time series representation technique that starts by reducing the dimensionality of a given time series using its local extrema. Next, it physically separates the minima and maxima points for more intuitiveness and consistency of the so-obtained time series representation. The second step consists in adapting the Dynamic Time Warping (DTW) measure so as to evaluate the score of similarity between the generated representations. We test the performance of LE-DTW on a wide range of real-world problems from the UCR time series archive for TSC. Experimental results indicate that for short time series, the proposed method achieves reasonable classification accuracy as compared to DTW. However, for long time series, LE-DTW performs much better. Indeed, it outperforms DTW while providing competitive results against popular distance-based classifiers. Moreover, in terms of efficiency, LE-DTW is orders of magnitude faster than DTW.

29 citations