Dear Readers,

It gives me great pleasure to announce the fourth regular issue of 2026. In this issue, 6 papers by 25 authors from 6 countries - Colombia, Greece, Mexico, Republic of Serbia, South Africa, Türkiye - cover a great variety of topical aspects of computer science.

In a continuous effort to further strengthen our journal, I would like to expand the editorial board: If you are a tenured associate professor or above with a strong publication record, you are welcome to apply to join our editorial board. We are also interested in high-quality proposals for special issues on new topics and trends.

As always, I would like to thank all the authors for their sound research and the editorial board and guest reviewers for their extremely valuable review effort and suggestions for improvement. I also want to thank the readers for their interest in our articles, which is reflected in the increasing number of accesses and PDF downloads.

In the fourth regular issue, I am very pleased to introduce the following 6 accepted articles:

Wicus J. van der Linden, Trienko L. Grobler, and Lynette van Zijl from South Africa investigate in their article how class imbalance and adverse imaging conditions affect optical Braille recognition (OBR), by comparing the current standard multiclass character‑based models with multilabel, dot‑based classification models, supported by a novel resampling strategy that exploits Braille symmetry. Empirical results confirm that the proposed framework achieves more robust and generalized performance under diverse conditions, positioning it as a new standard modelling framework for multilingual OBR systems.

Yelda Fırat, Yılmaz Kılıçaslan, Hüseyin Ali Sarıkaya, and Murat Kaan Yılmaz from Türkiye focus in their research on the need for scalable and objective early screening of Autism Spectrum Disorder (ASD) by applying a hybrid deep learning approach - integrating Convolutional Neural Networks, Bidirectional Long Short-Term Memory, and attention mechanisms - to detect motor biomarkers from children's natural home video recordings. The findings demonstrate that the model achieves over 97% accuracy on controlled datasets and over 83% on real-world public videos, offering a robust, clinically applicable screening tool that overcomes the limitations of artificial laboratory environments.

Sava Stanisic, Borislav Djordjevic, Branislav Belotic, Olga Ristic, Ivan Tot, Kristina Zivanovic, and Dimitrije Kolasinac from Republic of Serbia discuss in their article the challenge of minimizing end-to-end latency in dynamic, heterogeneous Kubernetes clusters by proposing a reinforcement learning-based orchestration framework that employs a deep Q-network (DQN) agent to make scheduling and migration decisions. The proposed approach achieves up to 25–32% reduction in average latency compared to default Kubernetes schedulers, demonstrating superior load balancing and effective adaptation to workload variability through learned placement policies.

In a collaborative effort between researchers from México and Colombia, Mirna Muñoz, Gabriel Garcia-Mireles, Jezreel Mejia, Yadira Quiñonez, Gloria Gasca-Hurtado, and Adriana Peña investigate in their research how Software Engineering (SE) provides methodological support throughout the Machine Learning (ML) development lifecycle. This study conducts a Systematic Mapping Study that examines current SE methodologies for ML-based systems and identifies gaps across lifecycle phases. The results indicate a significant focus on testing operations, with insufficient software engineering support for other phases, underscoring the need for more comprehensive frameworks, procedures, and standards to improve the stability, maintainability, and quality of machine learning-based software.

Rabia Tintin and Sait Can Yucebas from Türkiye report on the study addressing the limitations of existing sentiment analysis approaches for the Turkish language by proposing Duygu-Turk, a context-aware and linguistically enriched deep learning framework based on Plutchik’s Wheel of Emotions and non-monotonic logic. The results demonstrate that the proposed model significantly outperforms state-of-the-art transformer-based models by achieving high accuracy in both polarity and fine-grained multi-class emotion classification, highlighting its effectiveness for morphologically rich and low-resource languages.

And last but not least, Dimitrios Psilias, Athanasios Milidonis, and Ioannis Voyiatzis from Greece propose in their article an FPGA-based architecture for the combined secure transmission of UAVs' telemetry and high-definition video data using a single AES-128 module. Experimental results show that using this approach, a high throughput of 25.6 Gbps is achieved, without having significant overheads in execution delay and power consumption.

Enjoy Reading! 

Best wishes,

Christian Gütl, Managing Editor-in-Chief

Graz University of Technology, Graz, Austria