Faculty of Computer Science and Engineering

Permanent URI for this communityhttps://repository.ukim.mk/handle/20.500.12188/5

The Faculty of Computer Science and Engineering (FCSE) within UKIM is the largest and most prestigious faculty in the field of computer science and technologies in Macedonia, and among the largest faculties in that field in the region. The FCSE teaching staff consists of 50 professors and 30 associates. These include many “best in field” personnel, such as the most referenced scientists in Macedonia and the most influential professors in the ICT industry in the Republic of Macedonia.

Browse

Search Results

Now showing 1 - 10 of 434

Some of the metrics are blocked by your
consent settings
Multimodal Deep Learning for Online Meme Classification
(IEEE, 2024-12-15)
Han, Stephanie
;
Leal-Arenas, Sebastian
;
;
Cavalcante, Charles C
;
Boukouvalas, Zois
Memes possess a humorous intent, yet they can also be used for malicious purposes. Analysing meme data has the potential to enhance content monitoring, identify emerging topics, and support content moderation in online platforms. Memes also represent an interesting use case for multimodal machine learning, as they combine text and image data. In this study, we explored the linguistic characteristics and analysed the convergent themes of five meme classes through common word extraction. Moreover, we compared the effectiveness of various machine learning models, i.e., unimodal (text or image) and multimodal (early fusion, late fusion) in binary and multiclass meme classification tasks. Our results on a large meme dataset showed that memes heavily adhered to current affairs, demonstrated by the high frequency of topical words across meme classes. Regarding model accuracy, early fusion achieved superior accuracy over late fusion in meme classification. Binary models outperformed multi-class classification methods. However, fusion models did not consistently surpass the accuracy of independent text or image-based models.
Some of the metrics are blocked by your
consent settings
IN THE FINE PRINT: INVESTIGATING EDTECH PROVIDERS’DATA PRIVACY COMMITMENT-TOOLS FOR SCHOOLS
(IATED, 2024)
Hillman, V
;
Barud, K
;
Henne, T
;
Saillant, C
;
Radkoff, E
In the evolving educational technology (edtech) landscape, quality assessment processes are integral to education governance and ensuring quality at all educational levels. Transparency in data processing provided to users and adherence to privacy laws by edtech providers have become critical concerns for building trust with education stakeholders. This study explores data protection practices of selected edtech providers using an innovative mixed-method approach combining manual assessments with Machine Learning techniques. Our research focuses on: (1) empirical analysis of the transparency and legality of the information shared with schools by providers based on the articulation of their data privacy policies (DPPs) (2) methodological exploration integrating human and ML-based analyses. These components scrutinize how edtech providers communicate their data processing practices to schools and comply with privacy regulations such as the General Data Protection Regulation (GDPR) and age-appropriate standards, outlined in their DPPs. These practices are crucial for building trust between schools and edtech providers and for updating relevant government policies that address the challenges of digitizing education (evidencing recent unethical and illegal data practices). Our motivation stems from the statutory requirements schools must meet to ensure they integrate quality edtech products into their operations. Conducting Data Privacy Impact Assessments and evaluating providers’ DPPs, as part of procurement, while protecting students’ basic rights, is costly, labor-intensive and requires expertise beyond pedagogy. Hence, our research focuses on seeking to develop a non-expert template that can streamline the initial assessments of DPPs and evaluate a provider’s transparency towards users; and, test innovative technologies to scale this demanding process effectively and efficiently. Initial findings were derived from the ML-supported assessment of 10 popular edtech providers’ DPPs. These findings highlight varying degrees of transparency and compliance with data protection requirements concerning data processing information for end-users. They also elucidate whether current ML techniques such as OpenAI’s chatGPT ensure reliable automated assessments or produce untrustworthy results. Our methodology evaluates the clarity and comprehensibility of DPPs through manual scrutiny and leverages ML techniques for analysis of large datasets. It identifies current errors associated with ML applications in this context. This dual approach enhances the robustness and scalability of our evaluation framework, offering insights on how future assessments of edtech could be standardized and automated. The study contributes to discussions at the intersection of education, technology, ethics, policy and governance, offering actionable insights for education stakeholders in navigating the complexities of data privacy regulation and promoting responsible edtech innovation. Our findings and methodology contribute to global discourse in education and research by addressing the datafication of education and the application of AI in legal and ethical assessment practices. We advocate for ethical edtech that not only enhances educational outcomes but also prioritizes transparency, legality, and ethical integrity, with assessment of ML tools which could support and facilitate schools’ procurement and assessment processes.
Some of the metrics are blocked by your
consent settings
A study on appropriate segment length for generalized cuff-less blood pressure estimation from ECG features
(IEEE, 2024-05-20)
Kuzmanov, Ivan
;
;
Lamenski, Petre
;
;
Madevska Bogdanova, Ana
Blood pressure (BP) refers to the pressure exerted on the blood vessels as blood travels through the body. Our ultimate goal is to build a stable model for BP estimation as part of a triage process. In this study, we experiment to determine a suitable signal segment only from electrocardiogram (ECG) signals, to ensure a fast and reliable process of the BP estimation. The used dataset contains only high-quality ECG and arterial blood pressure (ABP) signals extracted from the Medical Information Mart for Intensive Care, MIMIC II and MIMIC III databases by our methodology. It was processed three times using similar machine learning (ML) methodologies, with different segment lengths. Three different datasets are generated using a non-overlapping window with a size of 8, 15, and 30 seconds, with the same ECG features. Several linear and nonlinear Machine Learning models are built on these datasets, and their results are compared. Our best results were obtained by a light gradient-boosting machine (LightGBM) regression model trained on the 30-second dataset. The model achieves Mean Absolute Error (MAE) of 10.87, 6.55, and 7.29, and Root Mean Squared Error (RMSE) of 14.49, 8.68, and 9.68 for systolic blood pressure (SBP), diastolic blood pressure (DBP), and mean arterial pressure (MAP), respectively. The results of our experiment indicate that a duration of 30 seconds is the minimum length that provides informative features, fulfilling the need for real-time delivery.
Some of the metrics are blocked by your
consent settings
From linguistic linked data to big data
(2024-05-22)
;
Apostol, Elena-Simona
;
Garabík, Radovan
;
Gkirtzou, Katerina
;
Gromann, Dagmar
With advances in the field of Linked (Open) Data (LOD), language data on the LOD cloud has grown in number, size, and variety. With an increased volume and variety of language data, optimizations of methods for distributing, storing, and querying these data become more central. To this end, this position paper investigates use cases at the intersection of LLOD and Big Data, existing approaches to utilizing Big Data techniques within the context of linked data, and discusses the challenges and benefits of this union.
Some of the metrics are blocked by your
consent settings
ChatGPT in exam rooms: preliminary insights into student performance with and without AI assistance
(IATED, 2024)
Pesovski, Ivica
;
Santos, Ricardo
;
Henriques, Roberto
;
Trajkovik, Vladimir
Reactions to artificial intelligence, particularly ChatGPT in the field of education, have been varied all across the world. We watched institutions, cities, and even entire countries prevent it in the educational context, and we also experienced the opposite extreme, which was institutions supporting it and integrating AI into their routine operational workflow. However, what is the most effective method for students? Does having access to AI allow them to benefit from it, or does it hinder their ability to perform well in school? Our interest in answering these questions resulted with a study carried out with the participation of 65 college students who were studying computer science and engineering. Structured programming and object-oriented programming were the two types of programming classes that we looked into. Both were administered in the same manner and lasted for the same amount of time; however, whether or not students had access to ChatGPT throughout the examination was different. When taking tests in structured programming, they were permitted to use ChatGPT; but, when taking exams in object-oriented programming, they were specifically prohibited from using it. Within the scope of this study, the preliminary findings are provided. Results show that when students had subsequent exams under the same circumstances and with access to ChatGPT, they improved on the second exam by around 53% and displayed a strong positive correlation between the two subjects. However, restricting AI access on the second exam, led to a 39% decrease in results and more unpredictability in their outcomes.
Some of the metrics are blocked by your
consent settings
Predicting bootcamp success: using regression to leverage preparatory course data for tech career transitions
(IATED, 2024)
Santos, Ricardo
;
Pesovski, Ivica
;
Henriques, Roberto
;
Trajkovik, Vladimir
In our ever-evolving digital landscape, the demand for tech-savvy professionals is soaring. However, traditional education often falls short in equipping individuals with the practical skills needed by employers. Aspiring tech enthusiasts face a dilemma: they want to gain swift entry into the industry without committing to lengthy degree programs. Meanwhile, career changers seek streamlined paths to acquire relevant skills. Programming bootcamps provide a pragmatic solution. These intensive, short-term programs prioritize hands-on learning over theoretical depth. Participants emerge with coding abilities, web application development skills, and collaborative prowess — all within months. Bootcamps attract both young learners exploring alternatives to Bachelor's degrees and professionals switching careers to tech jobs. By bridging the education-employment gap, bootcamps empower individuals for junior roles in the tech sector. However, bootcamps also pose challenges. Many participants lack formal programming training, which can impact their bootcamp success. Institutions offering these programs are incentivized to create preparatory courses, ensuring fundamental skills and providing support mechanisms. In this study, we analyze the efforts of future boot campers in preparatory courses at a European university, using leave-one-out cross-validation on a dataset of 207 bootcampers to create a predictive regression model that uses information provided upon registration and their respective attendance at the preparatory courses. Then, we used this model to predict the final score of a new cohort of 58 students and measure the model's performance by measuring the mean, squared, and root mean squared errors on the test set. In the second step, we analyzed the importance of the variables used by the predictive model by measuring the R2 score and the relative tree-based feature importance for each variable. Our results show that data collected before the start of a bootcamp can be used to predict the success of a bootcamper as our Random Forest model predicted each participant's final grade with a mean absolute error of 17.67 points (grades vary between and 100). Moreover, our model explains 46% of the final grades' variability, with prior knowledge of the topic, level of instruction and the number of completed preparatory steps among the most relevant features. Implications for both research and practice are analyzed and discussed.
Some of the metrics are blocked by your
consent settings
AI-Powered Education: Rethinking the Way Programming is Taught Using AI Tools and Reversed Bloom's Taxonomy
(IEEE, 2024-11-06)
Pesovski, Ivica
;
Vorkel, Daniela
;
Trajkovik, Vladimir
Artificial intelligence (AI) is revolutionizing the way programming is taught and learnt and is redefining t he way programming skills are assessed in the continuously changing field of computer science education. This paper introduces a novel approach for teaching and evaluating programming knowledge by reversing Bloom's taxonomy in order to accommodate AI-powered learning. Traditionally, Bloom's taxonomy progresses from basic cognitive skills like remembering and understanding to advanced skills like creating. With AI tools such as WebSim.ai and Anthropic Artifacts now capable of generating sophisticated outputs, the focus shifts away from students' ability to create, as these tools handle that task effectively. Instead, this paper proposes placing a higher emphasis on students' ability to understand and critically analyze AI -generated solutions, assessing their comprehension and ability to reverse-engineer existing work. We call this the Reverse Bloom Taxonomy, where students begin with creation and then move toward deeper understanding of the subject matter. This paper outlines the methodology for applying this reversed framework in programming education and presents a promising concept for improving student learning outcomes. The discussion addresses challenges in implementation and emphasizes the strategic integration of AI -driven learning to prepare students for a technology-driven workforce.
Some of the metrics are blocked by your
consent settings
Benchmarking Parallel Electrocardiogram Compression Based on Successive Differences
(IEEE, 2024-11-26)
Shekerov, A
;
;
Gusev, Marjan
We focus on parallelization methods for an electrocardiogram data compression algorithm based on successive differences to gain insights into the advantages and disadvantages of parallel implementations. The experimental methodology exposes a comprehensive and systematic benchmarking process with varying input file sizes, hosting machine characteristics, and two popular parallelization approaches: OpenMP and MPI. We check the research hypothesis to see if parallelizing the compression algorithm can reduce the runtime while keeping the original algorithm’s compression results. Our analysis and discussion show that OpenMP outperforms MPI. An OpenMP implementation with 12 threads on a processor with six cores achieves the highest average speedup of 7 versus a single-thread implementation. Performance gains depend heavily on the utilized hardware and the degree of parallelism.
Some of the metrics are blocked by your
consent settings
Securing ai systems: A comprehensive overview of cryptographic techniques for enhanced confidentiality and integrity
(IEEE, 2024-06-11)
Cano Garcia, Jose Luis
;
Udechukwu, Izuchukwu Patrick
;
Bolaji Ibrahim, Isiaq
;
Chukwu, Ikechukwu John
;
Dağ, Hasan
The rapid evolution of artificial intelligence (AI) has introduced transformative changes across industries, accompanied by escalating security concerns. This paper contributes to the imperative need for robust security measures in AI systems based on the application of cryptographic techniques. This research analyzes AI-ML systems vulnerabilities and associated risks and identifies existing cryptographic methods that could constitute security measures to mitigate such risks. Information assets subject to cyberattacks are identified, such as training data and model parameters, followed by a description of existing encryption algorithms and a suggested approach to use a suitable technique, such as homomorphic encryption CKKS, along with digital signatures based on ECDSA to protect the digital assets through all the AI system life cycle. These methods aim to safeguard sensitive data, algorithms, and AI-generated content from unauthorized access and tampering. The outcome offers potential and practical solutions against privacy breaches, adversarial attacks, and misuse of AI-generated content. Ultimately, this work aspires to bolster public trust in AI technologies, fostering innovation in a secure and reliable AI-driven landscape.
Some of the metrics are blocked by your
consent settings
CYBERMACS ERASMUS MUNDUS PROJECT AS AN INNOVATIVE, INDUSTRY-FOCUSED INTERNATIONAL COLLABORATION ON CYBERSECURITY EDUCATION
(IATED, 2024)
Gücüyener, A
;
Dilan, E
;
;
;
Creutzburg, R
The dramatic rise in cyber-attacks has made cybersecurity a significant global concern. The cost of cybercrime to the world economy was about $1 trillion in 2020. Along with the increased demand for cybersecurity experts, the changing landscape of cybersecurity threats requires the urgent need for comprehensive, dynamic, and applied cybersecurity education to qualify cybersecurity professionals to prevent, mitigate, and manage cyber threats. In this paper, we propose an innovative teaching and learning approach that brings together academia and industry to address the growing talent gap in cybersecurity, called CyberMACS. CyberMACS, funded by Erasmus+ Mundus Framework, is a Joint Master's Degree Programme in "Applied Cybersecurity", offering a full-time 2-year European MSc programme to provide a solid cybersecurity background on educating future cybersecurity experts to detect, prevent, mitigate, and manage cyber-attacks. CyberMACS is a robust institutional and international collaboration operation for demonstrating European excellence in higher education with a high-level integrated and transnational study programme on applied cybersecurity targeting the best students worldwide. The innovative teaching and learning methodology of CyberMACS relies on four main dimensions: (i) close cooperation with industry, (ii) dynamicity in a curriculum focusing not only on complex technical competencies but also soft skills, (iii) collaborative working culture among three international Higher Education Institutions (HEIs), (iv) international mobility opportunities that are provided to students. In addition to the solid international mobility component, in coherence with the need analysis in the cybersecurity industry, the programme outcomes are designed to prepare students for a multidisciplinary career and for employment opportunities in cybersecurity companies, consulting firms, public/private companies that have cybersecurity/information security departments, including critical infrastructures operators such as power system, transportation systems, and financial technologies. Finally, with respect to the nature of the cybersecurity industry and rising needs, CyberMACS has a solid focus on consolidating the students' soft skills and teaching them how to think outside the box. In that sense, the Curriculum and the teaching methodology of the joint events (i.e. Summer Schools) are established to lead students to put into practice the following specific skills: critically analyse, synthesise, interpret and summarise complex scientific processes; ability to examine the technological, economic, human, legal, organisational, socio-political and policy challenges in the cybersecurity field; design strategies that integrate best practices in cybersecurity, regulatory compliance, and risk management; gain hands-on experience, developing and executing integrated strategies, policies, and safeguards to manage cybersecurity risks across any organisation. In line with this given background, in this paper, we aim to disseminate our innovative teaching methodology, best practices, and limitations for creating an international teaching collaboration on a dynamic and industry-focused discipline such as cybersecurity. While we aim to present CyberMACS' methodology to reach the best students worldwide, we intend to demonstrate how an international talent pipeline can be established and sustained in a field where the demand for skills constantly surges.

Faculty of Computer Science and Engineering

Browse

Filters

Settings

Sort By

Results per page

Search Results