Ali Seman

Dr. Ali Seman

orcid

Research Focus & Metrics

Academic Career Length

17Years (2010-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
29 August 2023 in IEEE Access
Verified
Journal Article
Open access
JEJafhate Edward
Jafhate Edward
MRMarshima Mohd Rosli
Marshima Mohd Rosli
ASAli Seman
Ali Seman
Class imbalance exists in many data domains, posing numerous challenges to the data research community. Medical datasets, in most cases are predominantly imbalanced in nature. Through tackling multi-class issues, most researchers preferred the conventional method of decomposing it into binary classes for a more convenient solution. This method is not applicable for solving sensitive and crucial domains, such as medical data. Classifying medical datasets require all the classes to retain their form and maintain clinical validity. In this article, we develop a rebalancing framework for the multi-classification of imbalanced medical data using SCUT (SMOTE and Cluster-based Undersampling Technique) to rebalance the imbalanced class distribution, a feature selection method using a combination of SHapley Additive exPlanations (SHAP) and Recursive Feature Elimination (RFE), and DES-MI (Dynamic Ensemble Selection for multi-class) for improved multi classification performance. There are two novelties that contributes to the performance of our framework: improvised SCUT by implementing two clustering algorithms, and our proposed pool classifier selection for DES-MI. The performance of the proposed framework was compared with other state-of-the-art imbalanced framework using eight imbalanced datasets, each with varying degrees of imbalance. The experimental results indicate that our proposed framework performed better with average performance of 81.77%, 73.57%, and 75.87% in terms of Macro Average accuracy, extended G-mean, and Macro Average AUC, respectively. Our framework drastically increases the overall performances, owing to its ability to significantly handles the multi-class imbalance problem.

6 October 2012 in BMC Research Notes
Verified
Journal Article
Open access
PDF file
ASAli Seman
Ali Seman
ZBZainab Abu Bakar
Zainab Abu Bakar
MIMohamed Nizam Isa
Mohamed Nizam Isa
Y-Short Tandem Repeats (Y-STR) data consist of many similar and almost similar objects. This characteristic of Y-STR data causes two problems with partitioning: non-unique centroids and local minima problems. As a result, the existing partitioning algorithms produce poor clustering results.

1 June 2015 in OMICS: A Journal of Integrative Biology
Verified
Journal Article
Open access
ASAli Seman
Ali Seman
ASAzizian Mohd Sapawi
Azizian Mohd Sapawi
MSMohd Zaki Salleh
Mohd Zaki Salleh
Y-chromosome short tandem repeats (Y-STRs) are genetic markers with practical applications in human identification. However, where mass identification is required (e.g., in the aftermath of disasters with significant fatalities), the efficiency of the process could be improved with new statistical approaches. Clustering applications are relatively new tools for large-scale comparative genotyping, and the k -Approximate Modal Haplotype ( k -AMH), an efficient algorithm for clustering large-scale Y-STR data, represents a promising method for developing these tools. In this study we improved the k -AMH and produced three new algorithms: the N k -AMH I (including a new initial cluster center selection), the N k -AMH II (including a new dominant weighting value), and the N k -AMH III (combining I and II). The N k -AMH III was the superior algorithm, with mean clustering accuracy that increased in four out of six datasets and remained at 100% in the other two. Additionally, the N k -AMH III achieved a 2% higher overall mean clustering accuracy score than the k -AMH, as well as optimal accuracy for all datasets (0.84–1.00). With inclusion of the two new methods, the N k -AMH III produced an optimal solution for clustering Y-STR data; thus, the algorithm has potential for further development towards fully automatic clustering of any large-scale genotypic data.

1 February 2012 in Trends in Bioinformatics
Verified
Journal Article
Closed access
ASAli Seman
Ali Seman
ZBZainab Abu Bakar
Zainab Abu Bakar
MIMohamed Nizam Isa
Mohamed Nizam Isa
This paper reports on the experimental results of the k-modes-type algorithms for partitioning Y-Short Tandem Repeats (Y-STR) data. The results were based on the clustering accuracy scores of five hard and three soft k-modes-type algorithms. Six Y-Short Tandem Repeats data sets were used as a benchmark for the evaluation. The results clearly indicated that the soft k-modes-type clustering algorithms are the most reliable algorithms for partitioning Y-STR data.

1 March 2010
Verified
Conference Paper
Closed access
ASAli Seman
Ali Seman
ZBZainab Abu Bakar
Zainab Abu Bakar
ASAzizian Mohd. Sapawi
Azizian Mohd. Sapawi
Centre-based clustering is among the most applicable method for partitioning objects into homogenous groups. This paper presents two Centre-based clustering; K-Means and K-Modes algorithms to investigate and evaluate the clustering results of Y-STR data. The main goal of this paper is to compare the accuracy of clustering Y-STR results for different types of data: numerical and categorical data. The results show that the Y-STR data is more favour to categorical data. The accuracy of the Y-STR, treated as categorical data is 49%, whereas the numerical data is only a 26% chance producing a good clustering result. However, the amount of time taken by numerical data is much better compared to categorical data.

1 June 2010
Verified
Conference Paper
Closed access
ASAli Seman
Ali Seman
ZBZainab Bakar
Zainab Bakar
ASAzizian Mohd. Sapawi
Azizian Mohd. Sapawi
This paper evaluates Y-STR Surname data for attribute value weighting in k-Modes clustering algorithm. Three categories weighting schemas: (1) Relative Value Frequency (RVF); (2) Uncommon Attribute Value Matches (UAVM); (3) Hybrid weighting schema are evaluated for Y-STR Surname data. The overall results show that the clustering accuracy of all methods produces in between 40–44% only. However, the idea of adapting a weighting schema still looks a promising method in order to improve the clustering accuracy for Y-STR data.