10th International Congress on Information and Communication Technology in concurrent with ICT Excellence Awards (ICICT 2025) will be held at London, United Kingdom | February 18 - 21 2025.
Sign up or log in to bookmark your favorites and sync them to your phone or calendar.
Authors - Jiahui Yu, Simon Fong, Jinan Fiaidhi, Sabah Mohammed, Richard Millham, Alexandre Lobo, Seon-Phil Jeong, Liansheng Liu Abstract - . Clinical pathways play a vital role in managing diabetes treatment, incorporating medical strategies customized to individual patient needs. In the case of insulin-dependent diabetes mellitus (IDDM), accurately determining insulin dosage and administration timing is essential for regulating blood glucose levels effectively. While general health guidelines exist, they lack personalization. The interactions between medication, lifestyle, and patient conditions are complex, with therapy patterns differing among patients. This paper introduces a combined data stream mining method with fuzzy rule generation to create realtime decision guidelines as clinical pathways for managing IDDM. These guidelines are derived from daily medication records and personal blood glucose trends, generated from continuously updated health data instead of relying on past records. Fuzzy rules are preferred for their ability to personalize treatment, handle complex interactions, adapt in real-time, and provide accurate, timely decisions. They allow for personalization by adapting to individual patient conditions, providing tailored insulin dosages and timings based on real-time data. This method effectively manages the complex interactions between medication, lifestyle, and patient conditions, which can vary greatly among patients. Additionally, fuzzy rules are generated from continuously updated health patterns, ensuring that decision-making reflects the current state of the patient's health rather than relying on outdated historical data. This flexibility accommodates fluctuations in glucose levels due to various factors, making the approach more responsive to both short-term and long-term medical effects. Ultimately, using fresh data leads to more accurate and timely decisions, crucial for maintaining appropriate blood glucose levels in IDDM patients. A computer-based simulation is provided to assess the most appropriate data stream algorithms for this task.
Authors - Assia Boukhamla, Tamer Abderrahmane Lafia, Nabiha Azizi, Samir Brahim Belhaouari Abstract - The high prevalence of cardiovascular diseases (CVDs) worldwide requires accurate diagnostic imaging, particularly through magnetic resonance imaging (MRI). The framework includes preprocessing for region-of-interest segmentation via ViTs, followed by PTQ to reduce model size while maintaining segmentation accuracy Using a small calibration dataset, we apply PTQ to compress the ViT, significantly reducing storage requirements and latency without compromising precision. Experimental results indicate that Float16 quantization achieves an optimal balance between compression rate and segmentation accuracy, demonstrating the feasibility of ViTs for real-time applications.
Authors - Cristinel Gabriel RUSU, Simona MOLDOVANU, Nilanjan DEY, Luminita MORARU Abstract - We are interested in exploring different visual patterns by training ma-chine learning (ML) classifiers on raw and foreground images for MI detection, which has been less studied. In this work, we train machine learning classifiers on raw images (containing background lines) and clear images (containing just foreground/object with background lines removed). Two ECG record datasets containing normal (N) and myocardial infarction (MI) data are analysed via high-level features provided by standard 12-lead ECG signals. Only the limb lead I was cropped from the 12-lead signals to generate the input data. Data augmenta-tion was used for a balanced dataset to prevent overfitting while maintaining the required spatiotemporal invariances for a correct diagnosis. The newly generated ‘clear’ dataset results show that the proposed model achieves high classification performance for the AD, KNN, and RF models, with accuracies that are 32.1%, 27.3%, and 18.5% higher than those of their ‘raw’ counterparts, respectively. These results prove the robustness of the model.
Authors - Volodymyr Kulivnuk, Oleksandr Hladkyi, Alexander Gertsiy, Tetiana Tkachenko, Tetiana Shparaga, Tetiana Mykhailenko, Ihor Vynnychenko, Kateryna Postovitenko, Rostislav Semeniuk Abstract - The development of natural and artificial information systems (IS) in medicine and health tourism is explored. The essence of information resources (IR) and information processes (IP) and their role in medical treatment and health tourism services is investigated. The structural model of the body information resources is substantiated. The information processes occurring in the human body are described. The informaciology model of the information accumulation in the human memory is proposed. The structural model of building principles of human body functional systems (FS) as well as the informaciology model of the human body FS are systemized. The informaciology model of formation of adap-tive results of the human activity is proposed. The natural and artificial infor-mation systems usage in medicine and health tourism is substantiated. The struc-tural models of the natural and artificial information systems are observed. The informaciology resources of artificial IS are explored. The structure of informaci-ology technologies in artificial IS is defined. The structural models of cybernetic systems and artificial (preformed) therapeutic ones are determined.
Authors - Niyantra Mohan Babu, K.Vijayan, Alekhya Devi Malepati Abstract - Autism spectrum disorder (ASD) is a developmental condition that affects social communication and behavioral intelligence. Many live out their lives in ignorance due to misdiagnosis or a lack of awareness. Individuals with high-functioning autism or Asperger’s Syndrome are hard to detect using a single method of detection because not every child will show the same symptoms. For example, a child with Asperger’s Syndrome with good eye contact but without social communication skills would be hard to detect using only their gaze points. This paper explores a method for the diagnosis of ASD through the integration of three models: gaze point tracking, quantitative behavioral checklist, and image processing. The eye-tracking technology pinpoints the coordinates of the gaze points and connects them to the training to detect whether the subject has ASD. Studies on this subject have shown discrepancies between detected and actual individuals with autism, as not all autistic children have an irregular gaze. (Yaneva et al. 2020) Insights into the child’s behavioral patterns are offered through Q-CHAT, which quantitatively categorizes the actions of the child. However, the Q-CHAT checklist is one-dimensional and differs in result as the child grows. (Howard et al. 2022) A pre-trained CNN VGG 16 model identifies traits in the children’s facial features as for the image processing model. But, not all individuals with autism have a facial structure that belies their disorder. (Anitha et al. 2024) This paper addresses all the problems through the novel integrated approach of ASD detection using multiple methods.
Authors - Vasileios E. Papageorgiou, Dimitrios-Panagiotis Papageorgiou, Georgios Petmezas, Pan-telis Dogoulis, Nicos Maglaveras, George Tsaklidis Abstract - This study presents a computationally efficient Convolutional Neural Network (CNN) enhanced with transfer learning for medical image classifica-tion. The method was rigorously tested on 3 tumor datasets: brain MRI, and lung and kidney CT scans. It leverages a pre-trained CNN on brain MRI images, fine-tuned with minimal re-training for the CT scans, achieving high classification accuracy. Transfer learning allows the model to adapt to cancer-specific features by utilizing insights from large datasets. Re-training on each tumor type using only 20 epochs, can deliver significant classification performance, demonstrating the method's efficiency. The CNN's computational efficiency ensures it is both accurate and scalable, making it suitable for use in resource-constrained environ-ments. This research highlights the potential of low-complexity deep learning (DL) to accelerate cancer diagnosis while balancing accuracy and efficiency. It shows that complex deep learning models are not always necessary, and optimal performance can be achieved with lower computational costs.
Authors - Jorge Lituma, Anthony Moya, Remigio Hurtado Abstract - Dementia, a critical global health challenge recognized by the World Health Organization (WHO), affects millions of lives, with more than 50 million cases reported in 2019, a figure projected to double by 2050. Among its forms, Alzheimer’s disease is the most prevalent, underscoring the urgent need for early detection to improve patient outcomes and mitigate societal impact. Leveraging recent advancements in artificial intelligence, this study introduces an innovative deep learning framework aimed at revolutionizing the diagnostic process, providing valuable insights for the scientific community and practical tools for medical professionals. The proposed approach is structured into five key phases: data collection, preprocessing, model training using transfer learning, quality metrics validation including Accuracy, Precision, Recall, and F1-Score—and result interpretation through integrated gradients. A robust dataset of over 40,000 MRI images was utilized, achieving an exceptional accuracy of 99.86% in classifying the stages of Alzheimer’s disease. To ensure interpretability, integrated gradients were employed to highlight critical neuroanatomical markers, such as cortical atrophy and enlarged ventricles, distinguishing patients with dementia from healthy individuals. These findings validate the model’s reliability and demonstrate its potential as an innovative tool for advancing Alzheimer’s diagnosis and care.
Authors - Douglas Amobi Amoke, Yichun Li, Syed Mohsen Naqvi Abstract - Adopting machine learning solutions for monitoring vessel behaviour and surveillance in the maritime domain shows excellent promise. However, significant challenges arise due to the lack of publicly available vessel trajectory data labelled with Automatic Identification System (AIS) information. A new automated system has been proposed to preprocess and label vessel trajectory data collected from AIS at the Port of New York (NY), Blyth Port in Newcastle (NCL), United Kingdom, and a combined dataset called NYCL to address the labelling problem. This automated labelling system functions in three key stages. The first stage involves data collection and processing. The second stage transforms raw AIS data into meaningful vessel trajectory information. The third stage annotates and labels these trajectories, concluding with classification. The processed AIS data create vessel trajectories, with labels automatically generated. Finally, this work explores the classification models to demonstrate the effectiveness of labelled vessel trajectories in various maritime tasks.
Authors - MS Hasibuan, R Rizal Isnanto, Suryatiningsih, Chae Min A, Lee Kyung Min, Park So Hyeong Abstract - This study aims to design and implement a waste bank application to improve waste management efficiency through digital solutions. The application provides a dashboard to track waste collection activities in real-time, displaying data on waste amounts, schedules, and user contributions, enhancing transparency and efficiency. Test results show the system improves waste bank operations by 25% and simplifies waste management reporting.
Authors - Levyta Farah, Nurul Sukma Lestari, Dendy Rosman, Dewi Andriani Abstract - MSMEs (Micro, Small, and Medium Enterprises) and tourism have a very close relationship and support each other. The collaboration between the two has great potential in improving the economy and regional development. Therefore, active collaboration is needed between tourist destinations and MSMEs in the regions to support each other and enhance the quality of tourism in Indonesia. This research investigates the influence of digital innovation and sustainable strategies on MSME performance with the Penta helix as a moderating variable. The population of this research is MSMEs in Tangerang City, with a sample size of 303 respondents. The results of this research are that digital innovation does not affect MSME performance, while sustainability strategy and Penta Helix have a positive effect on MSME performance. This research also shows that Penta Helix can moderate digital innovation and sustainability strategies on performance. This research clarifies the contribution of variables to the growth and sustainability of MSMEs, strengthens their position in the global market, and enables the development of more robust policies and business practices, potentially significantly contributing to overall economic growth and supporting tourism in the Tangerang area.
Authors - Anamika Dhawan, Pankaj Mudholkar Abstract - Precision Agriculture has put in a lot of enhancement in improving agriculture in the last two decades. Plant monitoring is one of the essential applications of Precision Agriculture. In this study, an IoT-based system for rice leaf disease detection that runs on solar power and makes use of integrated machine learning on a Raspberry Pi 4 Model B is presented. In the classification of two important rice diseases, bacterial leaf blight and rice blast, the built custom Convolutional Neural Network (CNN) model, which was translated to TensorFlow Lite (TFLite) format for edge deployment, obtained a remarkable 94.28% accuracy. For scalable, effective disease detection in rice farming, this solar-powered, cost-effective device integrates edge AI and IoT.
Authors - Luka Jovanovic, Aleksandar Petrovic, Milan Tuba, Miodrag Zivkovic, Eva Tuba, Nebojsa Bacanin Abstract - Strong security measures are required due to the growing use of IoT devices and constantly growing network sizes. In order to tackle some of the most important issues in IoT security, this paper investigates the use of optimization metaheuristics in XGBoost hyperparameter tuning. In particular, we suggest a brand-new modified metaheuristic algorithm that is intended to improve diversity throughout the search process and is modeled after the firefly algorithm (FA). Experiments with simulations on a newly released IoT security dataset show how well the proposed optimizer works to enhance model performance. While tackling important issues related to hyperparameter optimization, such as striking a balance between exploration and exploitation, the method achieves a noteworthy accuracy of 0.996853. These findings demonstrate how the suggested approach may strengthen network security by using more accurate predictive modeling, opening the door for scalable and effective IoT systems in progressively complex settings.
Authors - Deborah Prasetya Kusuma, Honey Paramitha Soetioso, Nurul Sukma Lestari Abstract - The aim is to examine how employee engagement influences performance, considering the roles of technology and work-life balance. Furthermore, this research also evaluates job engagement as a mediating variable between digital engagement and work-life balance on job performance. This study utilizes quantitative methods, gathering data through both online and in-person questionnaire surveys. The data analyzed using partial least squares structural equation modeling (PLS SEM) and Smart PLS software. The participants are Generation Z hotel employees in Jakarta, such as contract or permanent staff, daily workers, and part-timers who are influenced by technology, referred to digital engagement. A total of 240 respondents successfully completed the survey. The results are digital engagement has influence on job performance, work life balance has not significant influence on job performance, digital engagement, work-life balance, mediated by job engagement has influence on job performance. This research presents a novel conceptual framework for analyzing hotel performance. It also provides valuable insights for hotel management to develop strategies that enhance generation z employees’ performance by improving digital engagement and work-life balance while simultaneously supporting the hotel’s sustainability. For further research can examine variables that were not included in this study, such as digital addiction, job stress, management support, job environment, and motivation with a broader reach local hotel or comparing even until international.
Authors - Rayner Henry Pailus, Rayner Alfred Abstract - Pose face recognition systems often struggle with the variability of illumination and face poses, especially when images are captured in uncontrolled environments. This paper addresses these challenges by proposing a novel face recognition approach: Multiple Adaptive Derivative Face Recognition (MADFR). Our method focuses on optimizing face recognition at every processing level to enhance overall accuracy. By incorporating multiple illumination training samples and diverse training data, including both controlled and wild images, our approach improves the robustness of face recognition models. Our analysis highlights the limitations of existing models like FaceNet, particularly in handling images with multiple face poses and varying background illuminations. We propose pose estimation landmarking and localization with multiple landmarks, which significantly enhances discriminant features. The effectiveness of our approach is demonstrated through extensive experiments on three datasets: LFW, Pointing 04, and Carl Dataset. Our results show that the proposed MADFR system, combined with the ensemble method MADBOOST, consistently outperforms other models. Specifically, MFRF 10 emerged as the top-performing model across all datasets, exhibiting high accuracy and low error rates. This research makes a significant contribution to the eld of face recognition by providing a robust solution that effectively handles the complexities of real-world scenarios. In conclusion, the MADFR system, with its optimized processing and decision-making capabilities, demonstrates substantial improvements in face recognition accuracy, paving the way for more reliable and effective face recognition technologies.
Authors - Sayaka Matsumoto, Kunihiko Takamatsu, Shotaro Imai, Tsunenori Inakura, Masao Mori Abstract - In the context of higher education, Institutional Research (IR) has increasingly emphasized the use of data-driven tools such as student surveys to enhance educational practices and university operations. This study addresses challenges in managing and improving student surveys through advanced visualization techniques. We propose a third visualization method—a stacked bar graph—alongside two existing methods, the heatmap and bar graph with line overlay. This third method visually represents the progression of respondent dropout across questions, offering a detailed view of response continuity. The three visualization methods were used to compare pre- and post-improvement survey data, highlighting key factors such as question design and response behavior. The results indicate that reducing the number of questions and providing clear instructions significantly improve response rates, especially in the later sections of the surveys. The third visualization method effectively highlights these improvements by enabling precise monitoring of dropout trends and response continuity. This study situates its contributions within the interdisciplinary framework of Eduinformatics, integrating education and informatics to optimize educational processes. The proposed visualization methods offer practical tools for evaluating the quality of student surveys and ensuring the validity of collected data. While primarily aimed at student surveys, these methods have broader applicability to other survey-based research contexts.
Authors - Asmaa Berdigh, Kenza Oufaska, Khalid El Yassini Abstract - This study proposes a gradual transition from cabled to wireless communication in vehicles as a means of reducing weight and meeting regulatory requirements related to CO2 emissions, maintenance costs, and time to market. However, the study recognizes that different network domains and compartments in the vehicle have varying requirements and constraints. Therefore, a hybrid architecture between classical wired and wireless networks using Ultra-Wideband (UWB) was proposed as a starting point for testing the feasibility and obtaining feedback. We selected the Headlamp Control Module (HCM) as an application domain since it represents a reduced network consisting of a microcontroller unit (MCU) that operates as a slave to another electronic control unit (ECU) and sensors. This allowed the study to apply the proposed approach to a representative unit scenario. The study outlines the system architectural description for the selected system, the HCM. It describes the Controller Area Network (CAN) and UWB communication and analyzes the requirements that must be fulfilled to interchange both communication technologies. This paper proposes a CAN-UWB gateway system architecture and simulates it to evaluate its ability to meet communication requirements.
Authors - Eyad Mamdouh, Mohamed Gabr, Marvy Badr Monir Mansour, Amr Aboshousha, Wassim Alexan, Dina Reda El-Damak Abstract - This study presents an encryption algorithm for picture cubes that is based on complex differential equation-derived hyperchaotic systems. In order to enable efficient multidimensional encryption, the sensitivity to beginning conditions—a key component of chaos theory—has been extended into the hyperchaotic realm. The combination of DNA coding sequences with Linear Feedback Shift Registers (LFSRs) has increased the complexity of the method. The utilization of LFSRs provides secure pseudo-random sequences, whereas DNA coding adds more cryptographic depth. This combination has produced a strong encryption system that guarantees data security and resistance to sophisticated cryptanalysis attacks. The suggested encryption method has proven to be suitable for protecting volumetric picture data due to its superior performance in entropy, key sensitivity, and resilience to statistical attacks.
Authors - Berliana Tadjudin, Elencia, Davy Jivan Parmono, Tiurida Lily Anita Abstract - There are factors that indirectly influence consumer purchasing decisions in the restaurant industry. As consumer awareness toward environmental issues grows, the implementation of eco-friendly packaging, environmentally friendly visual appeal and adopting sustainable business model are a growing trend in the restaurant industry. In other hand, elements such as the aesthetics of the menu, food packaging, the design of the restaurant room, and the general brand awareness are also one of the factors that play an important role in influencing purchasing decisions, which could also be a factor toward consumer trust in the restaurants and loyalty toward the business. With both ideas in mind, this research was conducted to answer and analyze the impact of various elements on consumer buying decisions toward a restaurant adapting sustainability model. The research is conducted in Greater Jakarta Region and manages to gather 250 samples of respondents which are analyzed statistically, to investigate the validity of the hypothesis. The data gathered from the analysis shows that there are significant relationships between the variables.
Authors - Pablo Salamea, Remigio Hurtado, Rodolfo Bojorque Abstract - Recent advancements in deep learning have enabled the development of convolutional neural network (CNN) architectures, which have proven to be valuable tools in computer-aided diagnosis (CAD) systems. These systems assist radiologists in identifying regions of interest associated with pathologies in chest X-ray images, a diagnostic tool recognized as essential by the World Health Organization (WHO). The WHO highlights that chest X-rays are an accessible and cost-effective method, crucial for evaluating respiratory and thoracic diseases, particularly in resource-limited settings and during global health emergencies. In this study, the Vindr-CXR dataset was used, known for providing labeled chest X-ray images suitable for multi-label classification tasks. The process began with data preparation, where images and labels were grouped in a binary format and split into training and validation sets. Subsequently, pre-trained neural network architectures, such as VGG16, InceptionV3, ResNet50, and EfficientNetB0, were utilized with weights initialized from ImageNet. The initial layers of these architectures were frozen, and dense layers with sigmoid activation were added for multilabel classification. During training, the binary crossentropy loss function and the Adam optimizer were employed. The models were trained for a fixed number of epochs, with validation conducted at the end of each epoch to evaluate metrics such as accuracy and loss. Finally, predictions were generated on the validation set, and key metrics such as the ROC curve, precision, recall, and F1-Score were calculated. The models achieved a promising performance, with an accuracy of 0.72 in detecting thoracic pathologies. These findings highlight the potential of deep learning to enhance diagnostic precision and support clinical decision-making, reaffirming the critical role of chest X-rays
Authors - Pei-Jung Lin, Meng-Chian Wu,Jen-Wei Chang Abstract - To compare the effects of music and meditation on brainwave patterns and attention, this study designed a series of EEG-based experiments. Participants were instructed to either listen to music or engage in meditation, while their attention levels were assessed using a Rapid Serial Visual Presentation (RSVP) paradigm to validate brainwave differences under varying attentional states. EEG data were collected to analyze changes in attention during exposure to different types of music. Subsequently, mathematical computations were applied to quantify and summarize the pre- and post-intervention differences. The experimental results revealed significant differences in the impact of various music genres on attention. Listening to classical music effectively enhanced attention, whereas listening to popular music demonstrated a notable effect on emotional relaxation. Deep meditation yielded the greatest improvement in concentration, and its brainwave patterns closely resembled those observed when listening to classical music. An analysis of Arousal and Valence metrics indicated that meditation led to positive emotional changes. These findings suggest that both music and meditation can influence attention and emotional states.
Authors - John Khoo, Rayner Alfred, Khalifa Chekima, Rayner Pailus, Chin Kim On, Ervin Gubin Moung, Raymond Alfred, Oliver Valentine Eboy, Normah Awang Besar Raffie, Ashraf Osman Ibrahim, Nosius Luaran Abstract - Carbon stock serves as a crucial metric for assessing the quantity of carbon stored within terrestrial and aquatic ecosystems, exerting signi cant in uence on global carbon dynamics and climate change mitigation eorts. Eective management of carbon stocks is vital for regulating atmospheric carbon dioxide (CO2) levels and mitigating the adverse impacts of climate change. The study delves into the estimation of carbon stocks, particularly focusing on above-ground biomass (AGB) as a key component of carbon storage in forests. In addition, explores methods for estimating above-ground biomass (AGB) of carbon storage in forests. Traditional eld-based approaches, statistical methods like regression, and machine learning techniques such as deep learning oer varied strategies for AGB estimation. These methods leverage a variety of data to enhance accuracy and scalability. Through empirical examples, the study presents their eectiveness in informing conservation strategies and fostering sustainable development amidst environmental challenges.
Authors - Juan Dominguez, Carlos Carranco, Remigio Hurtado, Rodolfo Bojorque Abstract - Driver fatigue is one of the leading causes of road accidents worldwide, affecting concentration, reaction time, and vehicle control. Sleep deprivation, long driving hours, and monotonous conditions increase the risk, particularly among professional drivers and shift workers. Identifying early signs of fatigue is essential for improving road safety and preventing accidents. This study introduces a structured framework for detecting fatigue based on EEG and EOG signal analysis. Using the SEED-VIG dataset, the methodology integrates multiple stages, including data processing, feature selection, model training, and performance optimization. Various machine learning models were tested, with particular emphasis on Random Forest, LSTM networks, and ensemble techniques such as Gradient Boosting, XGBoost, and LightGBM. Additionally, explainability techniques like SHAP and LIME were applied to highlight critical fatigue indicators, such as variations in blink frequency, saccadic movements, and brainwave activity in the theta and delta frequency bands. Among the tested models, the optimized Random Forest approach yielded the highest accuracy, with an RMSE of 0.0257. These findings contribute to the advancement of fatigue monitoring technologies, offering practical solutions for real-time driver assessment and accident prevention.
Authors - Madhwendra Nath, Subodh Srivastava Abstract - Denoising of the heart sound signal is crucial part of the heart sound signal analysis, as it reduces the interfering noise such as respiration noise, gastric noise, speech, motion artifacts, and power-line interference from the signal. The In-band noise in a phonocardiogram (PCG) signal refers to noise or artifacts that overlap with the frequency range of interest for major heart sounds which is typically 20–100 Hz. To reduce this in-band noise; a Daubechies-wavelets based approach is proposed. The parameters of Dabuchies-wavelets are revamped. To judge the proficiency of the proposed method, a novel performance-metric-index, Noise-area-difference (NAD) has been introduced. It evaluates the Denoising performance. The proposed method is compared with three other existing methods. The comparison results reveal that the proposed method outperforms existing state-of-the-art Denoising of Heart sound signals.
Authors - Shahd Tarek, Ali Hamdi Abstract - Brain tumors represent one of the most critical health challenges due to their complexity and high mortality rates, necessitating early and precise diagnosis to improve patient outcomes. Traditional MRI interpretation methods rely heavily on manual analysis, which is timeconsuming, error-prone, and inconsistent. To address these limitations, this study introduces a novel deep attentional framework that integrates multiple Convolutional Neural Network (CNN) base models—EfficientNet- B0, ResNet50, and VGG16—within a Multi-Head Attention (MHA) mechanism for robust brain tumor classification. Convolutional features extracted from these CNNs are fed into the MHA as Query (Q), Key (K), and Value (V) inputs, enabling the model to focus on the most distinguishing features within MRI images. By leveraging complementary feature maps from diverse CNN architectures, the MHA mechanism generates more refined, attentive representations, significantly improving classification accuracy. The proposed approach classifies MRI images into four categories: pituitary tumor, meningioma, glioma, and no tumor. A dataset of 7,023 labeled MRI images was curated from public repositories, including Figshare, SARTAJ, and Br35H, with preprocessing steps to standardize dimensions and remove margins. Experimental results demonstrate the superior performance of individual CNNs—VGG16 achieving 97.25% accuracy, ResNet50 98.02%, and EfficientNet-B0 93.21%. Moreover, the ensemble model integrating VGG16, EfficientNet-B0, and ResNet50 achieves the highest accuracy of 98.70%, surpassing other ensemble configurations such as ResNet50 + VGG16 + EfficientNet-B0 (96.95%) and VGG16 + ResNet50 + EfficientNet-B0 (95.96%). These findings underscore the effectiveness of multi-level attention in refining predictions and provide a reliable, automated tool to assist radiologists. The proposed framework highlights the transformative potential of deep learning in medical imaging, streamlining clinical workflows, and enhancing healthcare outcomes.
Authors - Luis Puebla Rives, Connie Cofre-Morales, Miguel Rodriguez, Eduardo Puraivan, Marisela Vera Medinelli, Abigail Gonzalez, Ignacio Reyes, Karina Huencho-Iturra, Macarena Astudillo-Vasquez Abstract - This study analyzes the perception of both practicing and future English teachers regarding an activity designed under a didactic conceptual framework that uses SCRATCH as a tool to promote English language teaching to 4th grade primary school students. A survey was designed, validated by experts, and then applied to 28 participants. The reliability of the scale was analyzed, showing internal consistency of 0.96 and 0.99 using Cronbach’s alpha and G6, respectively. Implicative statistical analysis was used to explore the relationships between questions across different dimensions. The similarity tree identified two significant clusters with values of 0.6 and 0.54. The implicative graph and cohesive tree displayed implications with values exceeding 0.7. The findings highlight a high appreciation for the activity using SCRATCH, which is perceived as both viable and an effective facilitator of contextualized and meaningful learning.
Authors - La-or Kovavisaruch, Kriangkri Maneerat, Taweesak Sanpechuda, Krisada Chinda, Sodsai Wisadsud, Thitipong Wongsatho, Sambat Lim, Kamol Kaemarungsi, Tiwat Pongthavornkamol Abstract - The industrial sector in Thailand remains primarily characterized by traditional practices of Industry 2.0, which face significant challenges in transitioning to Industry 4.0. This research proposes a decentralized real-time location and status reporting system to address these issues. By utilizing Ultra-Wideband (UWB) technology combined with the Internet of Things (IoT), the newly developed "UWB Tag Plus" device eliminates the reliance on costly UWB gateways, instead transmitting data directly to cloud servers via 4G/5G networks. Implementing this system at an automotive parts assembly factory in Thailand reduced system costs by over 30%. The communications protocol between the tag and cloud server changed from IEEE 802.15.4 to TCP/IP, which enhanced operational flexibility. The proposed system makes advanced modernization more accessible for small and medium-sized enterprises. Furthermore, the "UNAI Data Analytic" tool provides real-time performance analytics for automated guided vehicles, empowering warehouse operators to optimize operations and improve efficiency.
Authors - Junichiro Ando, Satoshi Okada, Takuho Mitsunaga Abstract - Large Language Models (LLMs) like ChatGPT and Claude have demonstrated exceptional capabilities in content generation but remain vulnerable to adversarial jailbreak attacks that bypass safety mechanisms to output harmful content. This study introduces a novel jailbreak method targeting Autodefense, a multi-agent defense framework designed to detect and mitigate such attacks. By combining obfuscation techniques with the injection of harmless plaintext, our proposed method achieved a high jailbreak attack success rate (maximum value is 95.3%) across different obfuscation methods, which marks a significant increase compared to the ASR of 7.95% without our proposed method. Our experiments prove the effectiveness of our proposed method to bypass Autodefense system.
Authors - Hana Ulinnuha, Mukhlish Rasyidi, Yanti Tjong, Husna Putri Pertiwi, Wendy Purnama Tarigan, Michael Tegar Wicaksono Abstract - Recently, tourism villages are central to Indonesia’s tourism development strategy, contributing significantly to local, regional, and even national economic growth. With the increasing number of tourism villages, understanding tourists’ perspectives is essential for ensuring their sustainability. Tourist reviews on platforms provide valuable insights into their experiences and expectations. Sentiment analysis, widely used in tourism research, enables the extraction and identification of opinions from these unstructured data sources, offering a deeper understanding of visitor sentiments. This study employs Large Language Models (LLM) to analyze tourist reviews of Indonesian tourism villages. Unlike common methods, LLMs provide advanced capabilities for both sentiment analysis and the evaluation of the 4A tourism components—Attraction, Accessibility, Amenities, and Ancillary services. By examining positive, neutral, and negative reviews, the research identifies key factors that shape tourist experiences. The findings offer practical recommendations for tourism village managers, not only to enhances visitor satisfaction but also supports the government’s goal of fostering economic growth in tourism and rural areas. The study demonstrates the potential of LLM-based sentiment analysis as a valuable tool for advancing Indonesia's tourism industry.
Authors - Mark Bhunu, Timothy T Adeliyi Abstract - The proliferation of social media (SM) platforms has made them an integral part of our daily lives, significantly shaping how we interact and engage with the world. While SM offers benefits such as social connectedness and sup-port, its impact on the psychological health and well-being of young individuals has both positive and negative dimensions. Understanding these effects is essen-tial to developing strategies for mitigating the adverse outcomes associated with its use. A systematic literature review was conducted to explore the influence of SM usage on the mental health and well-being of young adults aged 18 to 35. Drawing insights from 25 publications across three databases, the study identified common themes related to SM's effects on this demographic. The findings reveal a correlation between SM use and mental health outcomes, with benefits includ-ing enhanced social support but also risks such as depression, anxiety, low self-esteem, and increased vulnerability to cyberbullying. These results highlight the urgent need for targeted interventions to address the negative consequences of SM on the mental health and overall well-being of young adults.
