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 - Ayoung YANG, Atsushi ISHIBASHI, Ryota IMAI, Tsuyoshi MIYASHITA, Tadasuke FURUYA Abstract - As interest in autonomous ship research grows and challenges from natural disasters increase, the accurate assessment of marine environments is becoming increasingly important. However, current marine environment assessments are primarily focused on evaluating marine resources and environmental conservation, with limited applicability to vessel navigation. This study proposes the creation of a 3D map that integrates both underwater and above-water data, specifically targeting key areas of vessel navigation. The above-water data were collected using LiDAR(Light Detection and Ranging), while the underwater data were mapped using multibeam sonar. This map offers a level of realism that is not achievable with traditional nautical charts, enhancing maritime safety and supporting the operation of autonomous ships through a new format of data.
Authors - Selwa ELFIRDOUSSI, Hind KABAILI, Ghita SEKKAT Abstract - The COVID-19 pandemic disrupted many sectors, including education. The confinement of administrative bodies, teachers, and students confronted us with an unavoidable reality: the need for distance learning. Once schools reopened, several countries and institutions began adopting blended learning models, combining both distance and face-to-face modes. This sudden shift revived research in the field of education, specifically what is known as "Educational Data Mining," a discipline aimed at developing new tools for extracting and utilizing educational data. This paper presents a Machine Learning Model aims to predict student performance in blended learning by understanding the impact of various social, economic, personal, and other factors on student performance, and to identify students at risk of failure.
Authors - Phuong Thao Nguyen Abstract - Artificial Intelligence (AI) plays a transformative role in modern auditing by revolutionizing traditional methodologies and enhancing the overall audit process. The integration of AI technologies in auditing allows for the analysis of vast amounts of financial data, enabling auditors to identify anomalies, trends, and potential errors with unprecedented speed and precision. The significance of AI in identifying financial errors is paramount, as it enhances the detection of discrepancies that may go unnoticed through conventional auditing practices. By leveraging advanced algorithms and machine learning techniques, AI can recognize patterns and flag unusual transactions, thereby significantly reducing the risk of financial misstatements. Moreover, AI enhances the accuracy, efficiency, and compliance of financial audits. Automated data processing and real-time analytics minimize manual intervention, allowing auditors to focus on higher-level analysis and judgment-based tasks. AI tools also facilitate continuous auditing, enabling organizations to maintain compliance with regulatory standards and improve overall financial reporting. This paper provides an overview of the innovative ways AI is reshaping the auditing landscape, emphasizing its potential to elevate the quality and reliability of financial audits while streamlining processes and reducing costs.
Authors - Asmaa Abdul-Razzaq Al-Qaisi, Maryam Yaseen Abdullah, Enas Muzaffer Jamel, Raghad K. Abdulhassan Abstract - New technologies, particularly in recent years, are revolutionising the way the world of cultural heritage, as well as museum and exhibition spaces, is understood. In this context, virtual reality (VR), in particular, is seen as a valuable tool to enrich and enhance traditional visits, using virtual elements to make visitors' experiences more engaging and interactive. Furthermore, as arousing emotions is a fundamental aspect in the creation of museum itineraries, VR techniques are flanked by physiological techniques such as electroencephalography (EEG) that allow for a comprehensive analysis of visitors' feelings. Using EEG-based indicators, this paper aims to analyse the emotional state of a sample of visitors engaged in a first physical, then virtual experience. Interaction, in this case, took place with five specially chosen objects belonging to the collection of the museum of handicrafts of Valle d’Aosta region in order to classify the different levels of involvement. The results suggest that EEG analysis contributed significantly to the understanding of emotional and cognitive processes in traditional and immersive experiences, highlighting the potential of VR technologies in enhancing participants' cognitive engagement.
Authors - Kannary Keth, Samia Ben Rajeb, Virak Han Abstract - This paper presents a comprehensive literature review of research articles on Building Information Modeling in the past decade in thirteen Asian countries, including Cambodia, Thailand, Vietnam, Lao, Indonesia, Malaysia, Philippines, Singapore, Brunei, and Myanmar. Based on a Scopus search using keywords such as Building Information Modeling /Modelling /Model /Management /BIM, barrier/challenge, and the names of the 13 countries, the review identified 81 journal articles. Thirty-two articles were selected to extract the barrier statements. Only literature from four countries, China, Vietnam, Indonesia, and Malaysia, was found and selected. The semantic analysis by NVivo software included word frequency based on the literature review. As a result, 45 main barriers with six classifications: Cost, Technology, People, Environment, Organization, and Education were identified. Furthermore, the classification with high potential factors to influence the adoption of BIM in those countries is the environment, which demonstrates the external concerns, including standards, legality, guidelines, and regulations. Moreover, the main concern in China is a need for more willingness and awareness of BIM; in Vietnam, there is a lack of national standards; in Indonesia and Malaysia, there is concern about high costs. However, the study’s limitations include limited literature sources, exclusion of non-English sources, exclusion of article citations, and absence of expert validation.
Authors - Ivan Ursul Abstract - This paper presents a comprehensive approach to real-time fall detection using advanced Transformer-based architectures tailored for deployment on resource-constrained devices. Our dataset, collected over four months using the WitMotion BWT61CL IMU and complemented by smartphone video recordings, provides a rich, multi-modal source for modelling fall and non-fall events in diverse environments. Our work focuses on the deployment and performance evaluation of three Transformer-based models—Standard Transformer, Performer, and Linformer— each optimized for latency and accuracy in processing timeseries accelerometer data. Rigorous data preprocessing, including noise filtering and feature extraction, was applied to enhance signal quality. We evaluate the models on a dataset comprising 403 samples, achieving a peak accuracy of 98% with the Standard Transformer, and competitive results of 96% with the Performer and Linformer. The Performer model emerges as the most efficient latency, achieving an average response time of 34ms, while the Standard Transformer and Linformer require 350ms and 110ms, respectively. This efficiency, combined with high sensitivity and specificity, underscores the Performer model’s suitability for real-time embedded systems. Our findings demonstrate that advanced Transformer models, with optimized hyperparameters and efficient architectures, can deliver accurate, low-latency fall detection solutions, paving the way for enhanced safety in applications requiring real-time monitoring on compact hardware.
Authors - Lillian Mzyece, Jackson Phiri, Mayumbo Nyirenda Abstract - Accurate rainfall forecasts are critical for various sectors, yet traditional methods struggle due to evolving and non-linear weather patterns. This study evaluates four machine learning algorithms—Support Vector Machines (SVM), Random Forest (RF), Neural Prophet (NP), and Long Short-Term Memory (LSTM)—to determine the most effective algorithm for rainfall forecasting in Zambia. Results show that Neural Prophet outperformed others, achieving the lowest RMSE (4.67), MAE (16.75), and MAPE (13.40%). Its autoregressive capabilities, interpretability, and reduced parameter complexity make Neural Prophet the preferred choice for forecasting rainfall trends in Zambia.
Authors - Nailah Al-Madi Abstract - The diagnosis of appendicitis is a challenge especially for children, as its symptoms overlap other diseases and children are unable to express their pain well. The misdiagnosis rate ranges from 28% to 57% in children. Machine learning is efficient in building models that can help predict diseases. XGBoost is one of the best machine learning models since it is based on ensemble learning approach. XGBoost has hyper-parameters that should be tuned well in order to achieve high performance. These parameters could be optimized to find the optimal or near optimal performance of XGBoost. In this paper, an Optimized- XGBoost model is proposed, which uses Genetic Algorithm to optimize seven parameters of XGBoost to achieve high performance. This Optimized-XGBoost is used to predict three class labels of pediatric Appendicitis, including diagnosis (appendicitis or no appendicitis), Severity (complicated or not complicated), and management(conservative or surgical). The experiments were implemented on Pediatric Appendicitis with 38 features and 780 records, and compared optimized-XGBoost with original XGBoost, and other well-known classifiers, such as DT, SVM, NB, KNN, RF, and Adaboost. Results show that optimized-XGBoost achieved highest results for accuracy, precision, recall and F1-Score. For example, the F1 score results for the prediction of severity is 96.15%, for the prediction of diagnosis is 99.36%, and for treatment is 99.36%.
Authors - Andre Viviers, Bertram Haskins, Reinhardt A Botha Abstract - Tracking gastropod chemical trails is time-consuming and error-prone. This paper argues that computer vision provides a viable alternative. Using selected image manipulation and segmentation techniques, an unlabeled dataset was generated. A simple K-Means clustering algorithm and manual labelling created a labelled dataset. Thereafter, a best-effort model was trained to detect gastropods within images using this dataset. Using the model, a prototype was created to locate gastropods in a video feed and draw trace lines based on their movement. Five evaluation runs serve to gauge the prototype’s effectiveness. Videos with varying properties from the original dataset were purposefully chosen for each run. The prototype’s trace lines were compared to the original dataset’s human-drawn pathways. The versatility of the prototype is demonstrated in the final evaluation by generating fine-grained trace lines post-processing. This enables the plot to be adjusted to different parameters based on the characteristics that the resulting plot should have. This research demonstrated that a gastropod tracking solution based on computer vision can alleviate human effort.
Authors - Maya Dimitrova, Nina Valchkova Abstract - The paper presents the concept of a ‘cyber-physical nurse’ from a feasibility perspective for wider inclusion in healthcare, in particular in relation to empathic communication with the patient. The results of a pilot study on user perception of two robotic and one human faces are presented and discussed in this context. Users attributed positive features to neutral agents’ facial expressions, but not negative, which increases the feasibility of introducing social robots in healthcare. Some guidelines for cyber-physical nurse design are discussed, addressing challenges to its possible implementation in hospitals, rehabilitation centers, and home care settings.
Authors - Tagnon Adechina Geoffroy Zannou, Semevo Arnaud Roland Martial Ahouandjinou, Manhougbe Probus Aymard Farel Kiki, Adote Francois-Xavier Ametepe Abstract - Sensor-based gas analysis has been the subject of much research, particularly in the development of electronic nose (e-nose) systems. E-noses are based on chemical sensors to detect and analyze volatile organic compounds, and thus play an important role in a variety of fields. In this paper, based on three research strings, we have performed a bibliometric analysis to examine current trends and scientific contributions in the field of sensors for detecting odors and volatile organic compounds (VOCs), their use in electronic nose systems, work to improve their performance and their optimization. Using the Scopus database and English-language documents published between 2014 and 2024, we identify the most prolific authors, countries and journals in these fields. After that, a short literature review provides a detailed overview of the strategies to improve e-noses selectivity and reduce their drift. The results of the bibliometric analysis show a growing intercontinental interest, with strong scientific activity in China, the United States, India and Italy, with a particularly strong focus on performance improvement and sensor optimization. The short survey reveals the existence of a wide range of gas sensors with their advantages and disadvantages, significant advances in improving the performance of sensors and electronic noses, as well as new challenges that deserve attention.
Authors - Adibah Alawiah Osman, Azwan Abdullah, Sharul Shahida Shakrein Safian, Nor Zawani Ibrahim, Nik Rozila Nik Mohd Masdek, Norhasimah Shaharuddin, Nur Athirah Sumardi Abstract - The process of using various data methods within one study to con-firm that the results are firmly supported by the predictions made is called triangulation. Several methodological debates have highlighted the limitations of quantitative research compared to qualitative research. This paper hunts to explore the triangulation research approach in the background of Islamic marketing at fully established Islamic banks in Malaysia. Islamic marketing of Islamic banks is defined as the application of Islamic banking knowledge, Islamic advertising ethics, and the augmentation of learning and instruction by Islamic bank employees in this study. The present research clarifies the basis links between the quantitative data of Islamic bank’s staffs at fully established Islamic banks and the qualitative insights of Islamic financial experts. The amalgamation of both qualitative and quantitative approaches in data collection and evaluation significantly improves the quality of the research outcomes. Further studies ought to explore the application of the triangulation method in other domains.
Authors - Swastika Das Abstract - This paper discusses the role of the UMANG platform in achieving the goal of addressing the digital divide in Indian e-governance. The UMANG platform aggregates nearly 2,000 services across sectors--health, education, and finance--onto a singular mobile-first platform for which it strives to make accessible, transparent, and efficient. Under the Digital India initiative, the UMANG platform was launched in 2017. Despite rapid digitalization in India, especially in cities, its rural pockets lag significantly in terms of internet usage penetration, marking only 37.3% in rural areas, respectively. The present research looks into how the same platform is trying to reduce that gap by providing services in 22 Indian languages, Assisted Mode for those without proper digital literacy, and real-time updates in the furtherance of tracking services. Citizen engagement in the right direction, UMANG has streamlined interactions, minimised bureaucratic delays, and created transparency. Problems, however, are still seated there-like limited digital literacy, security of data, and resistance from some government departments. Finally, the study concludes that with continuous integration enhancements in digital security and wider citizen participation, UMANG can transform governance in India, paving the way towards realising the vision of Digital India.
Authors - D. I. De Silva, S.V. Sangkavi, W. M. K. H. Wije-sundara, L. G. A. T. D. Wijerathne, L. H. Jayawardhane Abstract - This study introduces FlexiMind, an innovative mobile application designed to support children aged 6–10 with specific learning disorders, including dyslexia, dysgraphia, and dyscalculia. By integrating evidence-based instructional strategies and leveraging modern technologies, the application delivers an inclusive and interactive learning environment. The app comprises four core modules: Dyslexia Assessment, Tamil Letter Learning, Math Hands, and Word Recognition & Sentence Construction. These modules employ multisensory approaches, including real-time feedback, gesture-based learning, and machine learning algorithms, to enhance cognitive, linguistic, and mathematical skills. Preliminary findings highlight significant improvements in handwriting accuracy, letter recognition, phonemic awareness, and mathematical comprehension among children using FlexiMind. With its focus on Tamil language support and an adaptive design, FlexiMind addresses the unique needs of Tamil-speaking children while offering scalable solutions for broader educational contexts. This study underscores the potential of technology-driven tools in transforming learning experiences for children with specific learning disorders.
Authors - Arjun Mehra, Arti Devi, Ananya Sharma, Sahil Rana, Shivam Kumar, K V Uday, Varun Dutt Abstract - Virtual reality holds enormous potential for disaster preparedness; yet, little is known about how varying landslide risk levels and environmental elements (day vs night) impact people's physiological and psychological responses to these simulated catastrophes. By utilizing behavioral measures (Euclidean distance around collision, number of collisions, and velocity around collision), this study closes this gap by investigating stress and cognitive responses. Eighty volunteers were divided into four groups at random, and each group was exposed to a distinct set of landslide probabilities under various conditions: low likelihood during the day, high probability during the day, and high probability at night. The findings indicate that perceived risk significantly increased behavioral measurements, independent of time of day. These results demonstrate VR's capacity to improve cognitive engagement and equip participants to handle the psychological difficulties that arise in actual crisis scenarios.
Authors - Akriti Agarwal, Harshvardhan Singh Gahlaut, Annie Jain, Shalini L Abstract - Most common complication of Diabetic mellitus is Diabetic Retinopathy: It causes the lesion to occur upon the retina and affects vision if not diagnosed early it triggers blindness. Diabetic retinopathy should be treated by an early diagnosis to avoid irreversible loss of vision. In addition, the manual diagnosis by ophthalmologists is less efficient and can easily miss the smallest detail that, in some cases, may not be visible to naked human eyes compared to the computer-aided systems. This implies proposing an existing supervised learning strategy for detection of DR from retinal fundus images to a hybrid combination of both deep learning InceptionV3 and ResNet and a machine learning model, namely Random Forest and Support Vector Machine. The model architecture incorporates advanced neural networks fused with classifiers which is further tuned and added up with an attention mechanism ensuring robust and one of the most accurate classification model of DR and non-DR cases. The dataset comprises of 30,000 fundus images which is preprocessed and augmented to improve model performance, hence addressing class imbalance. Additionally, a front-end app with Grad-CAM analysis is developed to classify DR and Non-DR images and visualize where the model focuses during classification.
Authors - Nadia Liz Quispe Siancas, Julian Llanto Verde, Wilder Nina Choquehuayta Abstract - Semantic segmentation of buildings using optical satellite images and deep learning techniques is essential for urban planning and monitoring, especially in suburban areas. In this study, we focused on evaluating the performance of six deep learning models: DeepLabV3 MobileNetV3, DeepLabV3 ResNet50, FCN ResNet50, EfficientNet-B0, ResNet101, and UNET. The dataset was collected from the province of Mariscal C´aceres, specifically in the district of Juanju´ı, located in the department of San Mart´ın, situated in the northeast of Peru. Our analysis revealed varying levels of precision for each model: DeepLabV3 MobileNetV3 achieved 74.14%, DeepLabV3 ResNet50 reached 83.35%, FCN ResNet50 attained 83.56%, EfficientNet-B0 yielded 61.37%, ResNet 101 obtained 63.60%, and UNET demonstrated 74.54%. These results provide insights into the effectiveness of different deep learning architectures for semantic segmentation tasks in suburban environments.
Authors - Ambroise D. K. Houedjissin, Arnaud Ahouandjinou, Manhougbe Probus A. F. KIKI, Francois Xavier Ametepe, Kokou M. Assogba Abstract - Image motion deblurring is an important issue in computer vision applications which encounter challenges like motion blur caused by camera shake, fast motion or irregular deformation of agricultural living things during image acquisition. Images acquired by UAV-embedded cameras are often blurred and usually error-prone in precision agriculture. So, image deblurring in applications such as plant phenotyping recognition, crop pests and diseases detection or animal behavior analysis is a great challenge. The main purpose of this paper is to carry out both a bibliometric analysis to assess the current research trends on UAV image motion deblurring with a brief survey of the main image motion deblurring techniques in agriculture. So, we used the Scopus database and 2138 articles were retrieved. This dataset has then been analyzed using a bibliometric tool. According to results, the most impactful authors have 53 and 46 publications respectively. Remote Sensing is the most impactful journal with an h-index of 49 and 285 published articles whereas China is the country with the most impactful production and the most cited document, indicating its considerable influence in this area of research. Results from the short survey indicate that further research is needed to develop more robust and efficient motion deblurring techniques tailored to the specific challenges of UAV imagery in precision agriculture.
Authors - Eduardo Puraivan, Patricio Tapia, Miguel Rodriguez, Steffanie Kloss, Connie Cofre-Morales, Pablo Ormeno-Arriagada, Karina Huencho-Iturra Abstract - This study provides empirical evidence on the effectiveness of large language models (LLMs), particularly ChatGPT, for automating the identification and analysis of cognitive demand levels in reading comprehension assessment tasks, using Barret’s Taxonomy. The manual classification of these tasks, even for experienced teachers, poses challenges due to their complexity and the time required. To address this issue, a three-step methodology was developed: selection of reading comprehension activities, automatic classification by ChatGPT, and comparison with the classifications from a group of experts. The experiment included 25 questions based on four readings extracted from a fourth-grade teacher’s guide for primary education. The results showed variability in the agreement between ChatGPT’s classifications and those of the experts: 77% in Activity 1, 50% in Activity 2, 52% in Activity 3, and 67% in Activity 4. At the question level, agreements ranged from 0% to 100%, highlighting discrepancies even among the evaluators, which underscores the inherent subjectivity of the task. Despite these divergences, the results emphasize the potential of LLMs to streamline the classification of educational activities on a large scale and the need to continue refining these models to enhance their performance in more complex pedagogical tasks.
Authors - Larissa de Lima, Priscila Capriles, Nathan Oliveira Abstract - This paper explores the use of machine learning (ML) with various physical, chemical, and biological parameter combinations to predict water quality, focusing on the Water Quality Index (WQI). We assess the performance of several regression algorithms across five different data combinations and examine the impact of inference and class balancing techniques on model outcomes. Our analysis reveals that LightGBM achieved the highest accuracy in WQI regression at 93%. This research introduces a novel approach to calculatingWQI by automating the traditional manual and complex parameter collection and calculation process. By streamlining water quality monitoring, our ML-based method offers a more efficient and innovative solution. Additionally, the study provides practical insights into handling data scarcity and using statistical inference for skewed sampling distributions.
Authors - Atiqur Rehman, Karim Elia Fraoua, Amos David Abstract - Blockchain technology has the potential to revolutionize traditional financial systems by offering decentralized, secure, and transparent transaction processing. This research focuses on developing a blockchain-based investment platform that integrates the Practical Byzantine Fault Tolerance (PBFT) consensus algorithm. The platform addresses critical issues faced by traditional investment systems, such as security vulnerabilities, inefficiencies, and the lack of transparency. By incorporating smart contracts, the platform automates key investment processes such as order placement and settlement, significantly reducing reliance on intermediaries. The system is designed to process transactions in real-time, offering high throughput and low latency, ensuring a smooth user experience. Extensive testing, including unit testing, integration testing, and security testing, has been conducted to verify the platform’s performance, scalability, and robustness. Security measures such as end-to-end encryption and multi-factor authentication (MFA) further enhance the platform's reliability. While PBFT ensures fast and secure consensus, the scalability of the algorithm may present challenges as the platform grows. Future work will focus on optimizing the PBFT system, exploring hybrid blockchain models, and integrating the platform with external financial systems to extend its applicability. The research demonstrates that blockchain, when combined with PBFT, can create a secure, efficient, and scalable solution for managing investment transactions.
Authors - Amira Jemaa, Adnan Rashid, Sofiene Tahar Abstract - Explainable Artificial Intelligence (XAI) plays an important role in improving the transparency and reliability of complex machine learning models, especially in critical domains such as cybersecurity. Despite the prevalence of heuristic interpretation methods such as SHAP and LIME, these techniques often lack formal guarantees and may produce inconsistent local explanations. To fulfill this need, few tools have emerged that use formal methods to provide formal explanations. Among these, XReason uses a SAT solver to generate formal instance-level explanation for XGBoost models. In this paper, we extend the XReason tool to support LightGBM models as well as class-level explanations. Additionally, we implement a mechanism to generate and detect adversarial examples in XReason. We evaluate the efficiency and accuracy of our approach on the CICIDS-2017 dataset, a widely used benchmark for detecting network attacks.
Authors - Ayush Verma, Krisha Patel, Hardikkumar Jayswal, Nilesh Dubey, Dipika Damodar Abstract - Insurance fraud significantly undermines the financial stability of the insurance industry, resulting in billions of dollars lost annually due to fraudulent claims across sectors like healthcare, auto, and property insurance. This paper proposes a robust methodology for detecting insurance fraud through the strategic implementation of ensemble machine learning algorithms, specifically XGBoost and Random Forest. By analyzing extensive datasets that include policyholder demographics, claim histories, and risk factors, we develop predictive models that accurately identify fraudulent activities while minimizing false positives. The effectiveness of our approach is supported by a comprehensive literature review highlighting the performance of various machine learning models in fraud detection, as well as our application of preprocessing techniques and feature selection to enhance model accuracy. Our findings indicate that the integration of advanced AI and ML technologies can revolutionize fraud detection in the insurance sector, offering a more secure and efficient environment for both insurers and policyholders.
Authors - Honorato Ccalli Pacco Abstract - Mushrooms are important in human nutrition due to their nutritional value in terms of protein, vitamin and mineral content. The volume of mushroom cultivation is currently increasing. This research focuses in the modeling and simulation of temperature, humidity and irrigation time controlling in mushroom cultivation in a protected environment. Using fuzzy logic in an intelligent system that allows process control and the LabVIEW software that facilitates graphic programming by means of virtual instruments, the irrigation time program was obtained as an output variable or an input variable-dependent response (input variables were temperature and humidity) in the intelligent system. The result was a program that shows how to act in different situations of temperature and humidity in mushroom cultivation in a protected environment. The fuzzy logic program in LabVIEW allowed the simulation of the system in terms of irrigation time in mushroom cultivation in a protected environment to achieve the expected results. In experimental results it can be observed that at low temperatures (15 °C) and low humidity (35%) the irrigation time is an average value (44.03). With the high temperature (35°C) and high humidity (95%) in the protected environment, the irrigation time will be with a low value (22.32). And it could be simulated by varying the input variables.
