Publications

An Auditable Confidentiality Protocol for Blockchain Transactions

Applying Gamification in Portuguese Learning

With the rapid development of mobile technology, teaching and learning become location-less and time-less. The increasing popularity of mobile technologies provides more opportunities for educators and developers to create a wider range of educational tools. Gamification in the field of education is to motivate students to enjoy learning by applying game design. This project seeks to apply gamification techniques in Portuguese learning for Chinese learners using mobile technology. Gamification elements, progression and leaderboard are used to encourage and motivate students to achieve better results. Finally, an evaluation survey was conducted to verify the effectiveness of beginners' language learning and building their confidence. Further studies could design experimental studies and adopt more qualitative research methods. Some different individual characters of students, like confidence level and cognitive style, could be considered in the future research. An idea to increase the interests and motivation of Portuguese learning is worth studying.

Integrating consortium blockchain into edge server to defense against ransomware attack

An edge server acts as a data gateway in an IoT network between IoT devices and backend servers. If the edge server is under the ransomware attack, the server's operation would be interfered by locking its critical files, leading to a single point of failure in the IoT network. This paper proposes a blockchain solution, called Blockchain-enabled Recovery Service (BRS), to tackle the malware injection attack in edge servers. In BRS, edge servers are implemented by nodes in a blockchain network which allows IoT data distributed in edge servers over multiple data connections. It ensures data availability and consistency while improving the network throughput. A simulation test has been conducted to evaluate the performance of edge servers in the blockchain network. The result showed that the size of data record in a ledger is crucial when maximizing the efficiency of a ledger and minimizing data loss in ransomware attack are considered. Moreover, even though the blockchain network is under the attack, data availability and data recovery can be achieved.

Application of machine learning in higher education to predict students’ performance, learning engagement and self-efficacy: a systematic literature review

Purpose In recent years, studies have shown that machine learning significantly improves student performance and retention and reduces the risk of student dropout and withdrawal. However, there is a lack of empirical research reviews focusing on the application of machine learning to predict student performance in terms of learning engagement and self-efficacy and exploring their relationships. Hence, this paper conducts a systematic research review on the application of machine learning in higher education from an empirical research perspective. Design/methodology/approach This systematic review examines the application of machine learning (ML) in higher education, focusing on predicting student performance, engagement and self-efficacy. The review covers empirical studies from 2016 to 2024, utilizing a PRISMA framework to select 67 relevant articles from major databases. Findings The findings show that ML applications are widely researched and published in high-impact journals. The primary functions of ML in these studies include performance prediction, engagement analysis and self-efficacy assessment, employing various ML algorithms such as decision trees, random forests, support vector machines and neural networks. Ensemble learning algorithms generally outperform single algorithms regarding accuracy and other evaluation metrics. Common model evaluation metrics include accuracy, F1 score, recall and precision, with newer methods also being explored. Research limitations/implications First, empirical research literature was selected from only four renowned electronic journal databases, and the literature was limited to journal articles, with the latest review literature and conference papers published in the form of conference papers also excluded, which led to empirical research not obtaining the latest views of researchers in interdisciplinary fields. Second, this review focused mainly on the analysis of student grade prediction, learning engagement and self-efficacy and did not study students’ risk, dropout rates, retention rates or learning behaviors, which limited the scope of the literature review and the application field of machine learning algorithms. Finally, this article only conducted a systematic review of the application of machine learning algorithms in higher education and did not establish a metadata list or carry out metadata analysis. Originality/value The review highlights ML’s potential to enhance personalized education, early intervention and identifying at-risk students. Future research should improve prediction accuracy, explore new algorithms and address current study limitations, particularly the narrow focus on specific outcomes and lack of interdisciplinary perspectives.

Ad Hoc Multiple Source Routing

OGPA: a novel indicator hinting at the academic performance of Chinese vocational college students using stepwise regression

Studying the factors influencing the academic performance of higher education students has been a major focus of attention for research scholars across the world. However, empirical research literature specifically addressing this topic in the context of Chinese higher vocational colleges remains limited. This study aims to construct a novel indicator that hints at the academic performance of Chinese college students. Using approximately 2,270 records of students majoring in economics and management from the 2016 to 2020 academic years, this comprehensive study analyzed various aspects of their academic and extracurricular performance, encompassing 33 variables. These variables ranged from fundamental demographic characteristics to essential academic metrics like Gaokao Scores (GKs), College English Test 4 (CET4), academic achievements, participation and outcomes in skill competitions, student honors received, challenges faced, and involvement in volunteer services. This study found that (1) GKs, CET4, National Computer Rank Examination (NCRE), Total Scholarship (TSS), Outstanding Student (OS), Outstanding Student Leader (OSL), General Difficulty Student (GDS), and Volunteer Service Hours (VSHs) all show a significant positive correlation with academic performance and (2) age was negatively correlated with academic performance. In addition, the study also found that the effect of volunteer service on academic performance showed a folded relationship between “valley” and “peak.” These findings provide a source of ideas and empirical references for college education administrators and educational policymakers when revising or developing policy documents.

Forecasting the Temperature of BEV Battery Pack Based on Field Testing Data

Monitoring electric vehicles' battery situation and indicating the state of health is still challenging. Temperature is one of the critical factors determining battery degradation over time. We have collected more than 2.3 million discharging samples via a custom Internet of Thing device for more than one year to build a machine-learning model that can forecast the battery pack's average temperature in real-world driving. Our best Bi-LSTM model achieved the mean absolute error of 2.92 $$^\circ $$ C on test data and 1.7 $$^\circ $$ C on cross-validation for prediction of 10 min on the battery pack's temperature.

Overcoming the Sudden Conversion to Online Education During the COVID-19 Pandemic: A Case Study in Computing Education

This paper presented a case study on an agile shift from brick-and-mortar to online distance learning as a measure to sustain teaching and learning in higher education in response to the interruption and impact brought by the outbreak of Novel Coronavirus Pneumonia (COVID-19) to the society. This case study provided a practical example demonstrating possible practices that can be adopted to deliver different types of modules and carry out assessment under force majeure circumstances. We have applied several methods in module delivery and reflected on their strengths and weaknesses. Our approaches have taken into consideration the different needs of taught and non-taught modules, robustness of online assessment, and quick orientation for students who were unfamiliar with online education. The consequent effectiveness of teaching and learning was evaluated in terms of instructor participation and student participation. The findings indicated a relatively slow start due to the series of necessary adjustments to be made to encounter the shift. However, once the adaptation process was completed, an enhancing trend was observed in students and instructors' engagement in various online teaching and learning practices. Our study highlighted that pastoral support is essential to maintain the motivation and engagement of students. In addition, adjustments to module delivery and assessment methods are highly recommended.

Multi-level lag scheme significantly improves training efficiency in deep learning: a case study in air quality alert service over sub-tropical area

In environmental monitoring, deep learning models are used where we can either use past observations or extrapolated values with high uncertainty as input. The lag scheme is commonly applied during the modeling and construction process, in the application of deep learning models in multivariate time series prediction. For an adaptive feature engineering, an automated lag scheme is essential for improving the training efficiency. In multivariate time series (MTS) models, the predictive accuracy of artificial neural network ANN-type models can be improved by including more features. It is assumed that when processing a certain number of multivariate features, the timeliness and lag time of the inter-influencing between any pair of elements are different. This research aims to adopt an adaptive approach to solve it, namely, multi-level lag scheme. The research methods include literature review, searching for relevant technology frontiers, feasibility studies, selection and design solutions, modeling, data collection and pre-processing, experiments, evaluation, comprehensive analysis and conclusions. In proof of concept, we demonstrated a practical case of seasonal ANN type MTS model and public service on air quality. In terms of models, ANN type models were attempted with ARIMA as the comparing baseline. We used public data set of more than two base stations with pollution varying from low to high and including southern to northern district of a small city. Conclusions can be drawn from the analysis of multiple experimental results, proving that the proposed solution can effectively improve the training efficiency of the model. This is of great significance, so that most such models can be implemented to adaptively use lagged past measured data as input, instead of synchronously inputting future prediction values, which can greatly improve the practical application of the model in predictive ability.

Fully On-chain Cloud Storage DApp on the Internet Computer Protocol

This article explores the feasibility of developing cloud storage dApps on the Internet Computer (IC) blockchain, which can hold full-stack applications entirely on-chain and run at web speed. We analyze cloud storage services' security, stability, cost, and performance issues and propose a multi-canister system design scheme for building cloud storage dApps on the IC. Our quantitative tests demonstrate that the blockchain-secured data I/O operations on the IC can result in ten times time savings and up to a million times cost savings compared to those on Ethereum; deploying small-scale data storage applications on the IC can even be more resource-efficient than on traditional servers. We also discuss the trade-offs and decision-making considerations in our conclusion.

Secure address configuration in IPv6 networks

To secure the operations of address auto-configuration protocols in IPv6 networks, a solution called Secure Address Configuration for IPv6 (SAC6) is proposed in this paper. Unlike the previous solutions that mainly use the cryptographic approach, SAC6 eliminates the threats to configuration protocols for the network nodes by acting like a Neighbor Discovery Protocol agent. The major merit of SAC6 is that its operations are transparent to the network and do not require the modification of existing protocols. Therefore, it can be seamlessly deployed in existing IPv6 networks. To demonstrate the viability of SAC6, we implemented SAC6 as a kernel module in a Linux bridge. In our experiments, SAC6 have successfully blocked various kinds of spoofing configuration protocol messages and prevent the network nodes from being attacked during the address configuration.

Enhancing Computing Curriculum with Collaborative Engagement Model to Enrich Undergraduate Research Experience

Attracting and engaging computer science students to enhance their thinking skills are challenging tasks. In the past 10 years, the Computing Program at the Macao Polytechnic Institute encouraged our undergraduate students in doing research work, striving to enhance their mathematical and algorithmic thinking skills as well as to engage them in their study. We are convinced that these experiences change students' perception of computer science and how they can be part of the innovation engine throughout their career in science and technology. This paper discusses the MPI Collaborative Undergraduate Research Engagement model which consists of three main building blocks: team work and interpersonal skills, domain knowledge building, and technical communications. Besides computer science, we believe that our model can also be applied to other disciplines.

The Data Exchange Protocol over Multi-chain Blockchain Using Zero-Knowledge Proof

The implementation of blockchain technology is becoming popular among cyber-physical systems. However, the current solutions suffer from scalability and privacy issues. In this position paper, we leverage zero-knowledge proof and multichain technology to propose an efficient system for data transferring across different components. Each component may maintain a private chain storing its data, and the system acts as a relayer between different chains, in which multiple private chains are efficient for appending new data. Only encrypted data is transferred from a source chain to a destination chain. The relayer handles data transferring in two phases: send and receive, and the relayer keeps a Merkle tree of all sent data. In fact, it only transfers the data if the receiver can submit a valid zero-knowledge proof that proves the ownership of the data. The zero-knowledge proof discloses nothing but the statement is true; therefore it protects anonymity for the data owners. This system is secure and satisfies relevant properties such as ledger indistinguishability, transaction non-malleability, and matchability.

Self-recovery Service Securing Edge Server in IoT Network against Ransomware Attack

Digital Library

Pose Attention-Guided Paired-Images Generation for Visible-Infrared Person Re-Identification

A key challenge of visible-infrared person re-identification (VI-ReID) comes from the modality difference between visible and infrared images, which further causes large intra-person and small inter-person distances. Most existing methods design feature extractors and loss functions to bridge the modality gap. However, the unpaired-images constrain the VI-ReID model's ability to learn instance-level alignment features. Different from these methods, in this paper, we propose a pose attention-guided paired-images generation network (PAPG) from the standpoint of data augmentation. PAPG can generate cross-modality paired-images with shape and appearance consistency with the real image to perform instance-level feature alignment by minimizing the distances of every pair of images. Furthermore, our method alleviates data insufficient and reduces the risk of VI-ReID model overfitting. Comprehensive experiments conducted on two publicly available datasets validate the effectiveness and generalizability of PAPG. Especially, on the SYSU-MM01 dataset, our method accomplishes 7.76% and 5.87% gains in Rank-1 and mAP. The code is available at <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://github.com/qyhsxdx/PAPG.</uri>

LLM-Integrated Semantic Deep Learning Framework for Automated Floor Plan Analysis, Estimation and Design Guidance

The rapid digitization of the real estate and architectural design industries has created a high demand for automated tools capable of parsing 2D raster floor plans. Traditional manual measurement and visual inspection are not only time-consuming but also highly susceptible to human error. In this paper, we propose a comprehensive, end-to-end deep learning framework designed to automatically extract rich semantic information from unstructured 2D floor plan images and provide professional design guidance via Large Language Models (LLMs). Our integrated pipeline employs the state-of-the-art YOLOv8 object detection model to accurately localize and classify 18 distinct architectural symbols and furniture items (e.g., doors, windows, beds, cupboards). Simultaneously, a U-Net architecture with a ResNet34 encoder is utilized for the precise semantic segmentation of structural elements, specifically walls and interior room spaces. To translate pixel-level predictions into actionable real-world metrics, we introduce a robust area calculation algorithm based on user-defined reference scale calibration. Furthermore, to bridge the gap between raw geometric data and actionable architectural intelligence, we introduce an LLM-driven evaluation module utilizing a local Ollama deployment and a Retrieval-Augmented Generation (RAG) pipeline to assess design compliance and quality. To overcome the scarcity of annotated architectural datasets, we implement a systematic data augmentation strategy, expanding a core dataset of 101 manually annotated floor plans to 303 varied instances, thereby significantly enhancing model generalization. Experimental results indicate that our YOLOv8-based detection module achieves a mean Average Precision (mAP50) of 92.3%, while the U-Net segmentation module achieves a mean Intersection over Union (mIoU) of 95.71%. Furthermore, the integrated system is deployed as a user-friendly, interactive web application, acting as an intelligent architectural assistant and demonstrating its practical viability and high efficiency for real-world engineering and architectural applications.

Taxi Demand and Fare Prediction with Hybrid Models: Enhancing Efficiency and User Experience in City Transportation

An essential part of a city’s transportation infrastructure, taxis allow for regular encounters between drivers and customers. Nevertheless, there are issues with efficiency since there is an imbalance in the supply and demand for taxis. This study describes the creation of a platform that serves both customers and taxi drivers by offering immediate forecasts of demand and fare. Root mean squared error (RMSE) of 3.31 and a negative log-likelihood of −3.84, the long short-term memory recurrent neural network (LSTM-RNN) with the mixture density network (MDN) is employed to forecast taxi demand. The best RMSE of 3.24 is obtained for fare prediction via an ensemble learning model that integrates linear regression (LR), ridge regression (RR), and multilayer perceptron (MLP). To ensure peak performance, the models are systematically created, implemented, trained, and improved. By integrating these models into a web application interface, the taxi service system offers a better overall user experience, which improves urban mobility.

The Positive Effect of Attention Module in Few-Shot Learning for Plant Disease Recognition

Few-shot learning is good solution for plant disease recognition which can generalize to new categories by using few samples. However, the features extracted from few shots are limited. Attention is a technique for focusing on the significant features which can help to obtain better feature representation. In this work, we use a naive metric-based few-shot learning network as the baseline method, exploit the effect of different kinds of attention module: channel attention, spatial attention and hybrid attention. In experiments, we choose the representative modules of each attention category to show their effects in few-shot learning paradigm: SE, ASPP, CBAM and Triplet Attention. We conduct experiments with two data settings of PlantVillage, and illustrate the usage these attention modules in Residual Networks. The results indicate that the different attention modules can improve recognition accuracy to varying degrees. Attention can be used as effective improvement of feature representation under few-shot condition.