Digital Library
Today, most students are with their smartphone. In classroom, teachers often ask questions to evaluate the students' understanding on a topic. An easy-to-use interactive response system with a companion mobile app will be useful for improving the teaching and learning experience. There are some voting systems in the market. However, most of them are not convenient to use and configure, making students and teachers reluctant to use. The system proposed in this paper includes three major functions - voting/quiz, countdown, and real-time statistics. The voting/quiz function provides a simple interactive approach that can be used in any subject. The system is designed with simplicity in mind. Teachers will find it easy to create a voting or question, and students will find it easy to do a voting or answer a quiz. The students can login question or voting by scanning the corresponding Quick Response (QR) code or entering the Randomly Generated Code, without a complicated login system for the students.
The quality of translation work mainly depends on the understanding of the words in their domain. If machine translation can accurately translate the words in a domain in different languages, it can even avoid any human communication error. To achieve this, a high-quality bilingual corpus is crucial as they are always the basis of state-of-the-art machine translation system. However, it is complicated to construct the corpus with large amount of parallel data. In this paper, a new crawling architecture, called Hybrid Crawling Architecture (HCA), will be proposed, which efficiently and effectively collects parallel data from the Web for the bilingual corpus. HCA aims at targeted websites, which contains articles in at least two different languages. As it is a mixture of Focused crawling architecture and Parallel crawling architecture, HCA takes advantages over both architectures. In intensive experiments on crawling parallel data of relevance topics, HCA significantly outperforms Focused crawling architecture and Parallel crawling architecture for 30% and 200% respectively, in terms of quantity.
In this paper, we propose a Multiple Disjoint Path (MDP) routing protocol to maximize network throughput and minimize the protocol overheads in wireless ad hoc networks. MDP consists of two components: virtual source routing and core heuristic. Virtual source routing constructs virtual paths that do not suffer from scalability, privacy and efficiency problems caused by long paths for DSR in large networks. Core heuristic is to ensure the efficiency of request packet propagation in route discovery operation. Our simulation results reveal that MDP performs at a satisfactory level in dense networks in terms of connectivity and transmission efficiency. Network resources can be utilized so that network throughput can be significantly increased at a cost of minimal protocol overheads comparatively.
Digital Library
In recent years, the number of vehicles on the road is increasing rapidly. The traffic loading of the roads increases and the traffic slows down. If a car accident occurs or a car is illegally parked on the road, the traffic would become congested. To ease the problem, it is crucial to understand the traffic on roads in real-time before resolving the solution. This project will show a new method to visualize the traffic on the road, which is called Differential Timing Method (DTM). By checking the passing time when vehicles enter and exit a road segment, we compare the average passing time with the standard passing time. The traffic condition on the road can be determined. To check the time when a vehicle is seen, the Bluetooth technology is used. When the vehicle passes by a Bluetooth reader, it will be discovered and the discovery time will be recorded down. Thus, a traffic dataset can be built. To visualize the traffic dataset, the data will be aggregated and illustrated by bar chart. From the bar chart, the traffic condition of the road will be disclosed and indicated as congestion, slow moving, or free flow.
It is crucial to speed up the training process of multivariate deep learning models for forecasting time series data in a real-time adaptive computing service with automated feature engineering. Multivariate time series decomposition and recombining (MTS-DR) is proposed for this purpose with better accuracy. A proposed MTS-DR model was built to prove that not only the training time is shortened but also the error loss is slightly reduced. A case study is for demonstrating air quality forecasting in sub-tropical urban cities. Since MTS decomposition reduces complexity and makes the features to be explored easier, the speed of deep learning models as well as their accuracy are improved. The experiments show it is easier to train the trend component, and there is no need to train the seasonal component with zero MSE. All forecast results are visualized to show that the total training time has been shortened greatly and that the forecast is ideal for changing trends. The proposed method is also suitable for other time series MTS with seasonal oscillations since it was applied to the datasets of six different kinds of air pollutants individually. Thus, this proposed method has some commonality and could be applied to other datasets with obvious seasonality.
In this paper, we propose a Multiple Disjoint Path (MDP) routing protocol to maximize network throughput and minimize the protocol overheads in wireless ad hoc networks. MDP consists of two components: virtual source routing and core heuristic. Virtual source routing constructs virtual paths that do not suffer from scalability, privacy and efficiency problems caused by long paths for DSR in large networks. Core heuristic is to ensure the efficiency of request packet propagation in route discovery operation. Our simulation results reveal that MDP performs at a satisfactory level in dense networks in terms of connectivity and transmission efficiency. Network resources can be utilized so that network throughput can be significantly increased at a cost of minimal protocol overheads comparatively.
Accurate prediction of storm surges is crucial for mitigating the impact of extreme weather events. This paper introduces the Bidirectional Attention-based Long Short-Term Memory (LSTM) Storm Surge Architecture, BALSSA, addressing limitations in traditional physical models. By leveraging machine learning techniques and extensive historical and real-time data, BALSSA significantly enhances prediction accuracy. Utilizing a bidirectional attention-based LSTM framework, it captures complex, non-linear relationships and long-term dependencies, improving the accuracy of storm surge predictions. The enhanced model, D-BALSSA, further amplifies predictive capability through a doubled bidirectional attention-based structure. Training and evaluation involve a comprehensive dataset from over 70 typhoon incidents in Macao between 2017 and 2022. The results showcase the outstanding performance of BALSSA, delivering highly accurate storm surge forecasts with a lead time of up to 72 h. Notably, the model exhibits a low Mean Absolute Error (MAE) of 0.0287 m and Root Mean Squared Error (RMSE) of 0.0357 m, crucial indicators measuring the accuracy of storm surge predictions in water level anomalies. These metrics comprehensively evaluate the model’s accuracy within the specified timeframe, enabling timely evacuation and early warnings for effective disaster mitigation. An adaptive system, integrating real-time alerts, tropical cyclone (TC) chaser, and prospective visualizations of meteorological and tidal measurements, enhances BALSSA’s capabilities for improved storm surge prediction. Positioned as a comprehensive tool for risk management, BALSSA supports decision makers, civil protection agencies, and governments involved in disaster preparedness and response. By leveraging advanced machine learning techniques and extensive data, BALSSA enables precise and timely predictions, empowering coastal communities to proactively prepare and respond to extreme weather events. This enhanced accuracy strengthens the resilience of coastal communities and protects lives and infrastructure from the escalating threats of climate change.
Few-shot learning (FSL) is suitable for plant-disease recognition due to the shortage of data. However, the limitations of feature representation and the demanding generalization requirements are still pressing issues that need to be addressed. The recent studies reveal that the frequency representation contains rich patterns for image understanding. Given that most existing studies based on image classification have been conducted in the spatial domain, we introduce frequency representation into the FSL paradigm for plant-disease recognition. A discrete cosine transform module is designed for converting RGB color images to the frequency domain, and a learning-based frequency selection method is proposed to select informative frequencies. As a post-processing of feature vectors, a Gaussian-like calibration module is proposed to improve the generalization by aligning a skewed distribution with a Gaussian-like distribution. The two modules can be independent components ported to other networks. Extensive experiments are carried out to explore the configurations of the two modules. Our results show that the performance is much better in the frequency domain than in the spatial domain, and the Gaussian-like calibrator further improves the performance. The disease identification of the same plant and the cross-domain problem, which are critical to bring FSL to agricultural industry, are the research directions in the future.
Processing ancient text images presents significant challenges due to severe visual degradation, missing glyph structures, and various types of noise caused by aging. These issues are particularly prominent in Chinese historical documents and stone inscriptions, where diverse writing styles, multi-angle capturing, uneven lighting, and low contrast further hinder the performance of traditional OCR techniques. In this paper, we propose a unified neural framework, UniText, for the detection, recognition, and glyph restoration of Chinese characters in images of historical documents and inscriptions. UniText operates at the character level and processes full-page inputs, making it robust to multi-scale, multi-oriented, and noise-corrupted text. The model adopts a multi-task architecture that integrates spatial localization, semantic recognition, and visual restoration through stroke-aware supervision and multi-scale feature aggregation. Experimental results on our curated dataset of ancient Chinese texts demonstrate that UniText achieves a competitive performance in detection and recognition while producing visually faithful restorations under challenging conditions. This work provides a technically scalable and generalizable framework for image-based document analysis, with potential applications in historical document processing, digital archiving, and broader tasks in text image understanding.
A novel Neural Offset Min-Sum(NOMS) Belief Propagation(BP) decoding algorithm based on model-driven is proposed which applied to LDPC decoding. NOMS is improved multiplication in Neural Normalized Min-Sum(NNMS) into addition operation to reduce the complexity of calculation., a better Bit Error Rate (BER) performance is simultaneously achieved in the same condition. Secondly, considering that there are still many multiplication operations in NOMS, we propose a novel Shared Offset Min-Sum(SNOMS) to reduce the number of weights in the network by sharing parameters. Finally, codebook-based quantization is used to further reduce the memory consumption. Simulation experimental results show that the proposed method has a better BER performance, and the decoding accuracy of the decoder is 0.65dB higher than that of the NNMS after 5 iterations. In addition, SNOMS decoding method achieves almost the same decoding performance comparable to that of NOMS, but requires less complex calculation. Proposed quantization of code-book method reduces memory requirement significantly with slight performance loss.
This research intends to show how an analytical cyclic division of a dataset can improve ANN-type models in predicting future situation of different air pollutants in small-sized urban cities. Similar to other sub-tropical cities, the four seasons are not significant...
Blockchain exposes all users' transaction data to the public, including account balances, asset holdings, trading history, etc. Such data exposure leads to potential security and personal privacy risks that restrict blockchain from broader adoption. Although some existing projects focus on single-chain confidential payment, no existing cross-chain system supports private transactions yet, which is incompatible with privacy regulations such as GDPR. Also, current confidential payment systems require users to pay high extra fees. However, a private and anonymous protocol encrypting all transaction data raises concerns about malicious and illegal activities since the protocol is difficult to audit. We need to balance privacy and auditability in blockchain.We propose an auditable and affordable protocol for cross-chain and single-chain transactions. This protocol leverages zero-knowledge proofs to encrypt transactions and perform validation without disclosing sensitive users' data. To meet regulations, each auditor from an auditing committee will have an encrypted secret share of the transaction data. Auditors may view the private transaction data only if a majority of the committee agrees to decrypt the data. We employ a ZK-rollup scheme by processing multiple transactions in batches, which reduces private transaction costs of 90\% compared with solutions that do not employ the ZK-rollup. The proposed scheme has been implemented using Zokrates and Solidity and evaluated running on the Ethereum test network. The results show that the total one-to-one private transactions latency is about 5 seconds. Moreover, the security of the protocol is analyzed by mean of the standard real/ideal world paradigm.
As the number of vehicles running on the road increase, the traffic control becomes a serious problem in many historical cities, like Macao. Public transportation would hardly run and manage. In this research project, we propose the Electronic Vehicle Identification (EVI), which is a smart device with low-energy Bluetooth connection that exchanges the vehicle information and allows external communication with other control systems. As an ID for vehicle, it can be uniquely identified a vehicle and communication with vehicle is allowed. To illustrate the correctness of EVI, we conducted an experiment by using a Bluetooth reader to connect to EVI to discover the movement of bus vehicle departing or arriving at the bus terminal. The result shows that the bus vehicles can be captured on the road effectively, which can be used to support the Macao traffic department in optimizing road resources, managing traffic and making development plan.
