Journal of Computer and Forensic Sciences

Surface Electromyography-based Gesture Recognition for Robot Hand Control

Srđan Savić — 2024-12-25

This paper presents a human-machine interaction interface for a teleoperated control of a three-finger robot hand based on a surface electromyography. Based on the recorded sEMG signals, motion intention of a human operator is recognized and mapped onto corresponding predefined robot hand grasps. First, the experimental setup and underlying methodology of dataset generation are presented. Namely, 4-channel EMG data were collected from the forearm muscles of 4 healthy subjects as they performed a sequence of predefined grasps. Data processing included: data filtering and segmentation, feature extraction both in time and frequency domains, data annotation, and gestures classification. Two types of classifiers (with multiple variations of parameters) were implemented and compared: (i) Support Vector Machines with multiple different kernels and (ii) K-Nearest-Neighbors. The k-cross validation procedure was combined with the holdout method to generate training, validation and test sets. Performance of each classifier is presented in the form of a confusion matrix and a set of statistical measures, including F-measure. Finally, the integration of the proposed sEMG-based HMI interface and the selected classifier with a robot hand motion controller is presented.

Comparative Analysis of Certain Clustering Algorithms That Do Not Require Predefined Number of Clusters on the Articles of the Criminal Code of the Republic of Serbia

Sanja Raičević — 2024-12-25

This paper explores the application of certain clustering algorithms for analyzing textual documents from the Criminal Code (CC) of the Republic of Serbia (RS). Clustering was performed using three popular algorithms: DBSCAN, Mean-Shift, and Hierarchical Clustering (Agglomerative Clustering). The input data consisted of textual documents in .txt format, where each document corresponds to a specific article of the law. The aim of this study is to identify thematic groups within the legislative texts and analyze the advantages and disadvantages of each algorithm in the context of the specific characteristics of legal documents. The clustering results using these three algorithms show that DBSCAN faces challenges with noise, while Mean-Shift effectively detects dense clusters, and Hierarchical Clustering allows for a detailed analysis at various levels of granularity. In conclusion, this paper provides valuable insights into the application of these clustering algorithms to legal texts and offers recommendations for their selection when analyzing similar datasets.

The clustering approach to malware dataset analysis

Slaviša Ilić — 2024-12-27

The research in the area of malware analysis is very popular with accent on machine learning algorithms that help automate this subject. One of the leader portals that help researchers with dataset problems is Virus Total, providing free academic accounts with hundreds of thousands malware samples with metadata. This work contributes with the analysis of 429058 malware samples from Virus Total in terms of overcoming the problem of inconsistent labeling of the antivirus scan results from different vendors. Two methods were used, LSA and LDA both with automatic calibration of parameters, with purpose of finding the optimal number of clusters - both resulting with 5. The graphical representation of the clusters was done by k-menas clustering in two dimensional space. Additional research on most informative words in each cluster, showed that we can report of 4 similar classes, and one cluster per method (LSA and LDA) that was not related to the cluster in opposite method by wrd meaning in it. The showed results give good approach malware data analysis when dealing with inconsistently labeled dataset.