A5060707834- Optimization and Search Problems
- Topic Modeling
- Online Learning and Analytics
- Advanced Bandit Algorithms Research
- Natural Language Processing Techniques
- Online and Blended Learning
- Recommender Systems and Techniques
- Distributed systems and fault tolerance
- Educational Games and Gamification
- Sentiment Analysis and Opinion Mining
- Parallel Computing and Optimization Techniques
- Advanced Graph Neural Networks
- Advanced Database Systems and Queries
- Distributed and Parallel Computing Systems
- Virtual Reality Applications and Impacts
- Intelligent Tutoring Systems and Adaptive Learning
- Mobile Learning in Education
- Technology-Enhanced Education Studies
- Misinformation and Its Impacts
- Algorithms and Data Compression
- Augmented Reality Applications
- Complexity and Algorithms in Graphs
- Data Management and Algorithms
- Genomics and Phylogenetic Studies
- AI in Service Interactions
City University of Hong Kong
2020-2024
Hong Kong Polytechnic University
2022
Hong Kong Metropolitan University
2014-2021
Tokyo Metropolitan University
2021
Education University of Hong Kong
2021
Nihon Fukushi University
2021
Kasetsart University
2020
University of Hong Kong
2006-2014
Center for Massive Data Algorithmics
2011-2012
Aarhus University
2011-2012
To tackle the exponentially increasing throughput of Next-Generation Sequencing (NGS), most existing short-read aligners can be configured to favor speed in trade accuracy and sensitivity. SOAP3-dp, through leveraging computational power both CPU GPU with optimized algorithms, delivers high sensitivity simultaneously. Compared widely adopted including BWA, Bowtie2, SeqAlto, CUSHAW2, GEM GPU-based BarraCUDA CUSHAW, SOAP3-dp was found two tens times faster, while maintaining highest lowest...
In this paper, we give a simple scheme for identifying ε-approximate frequent items over sliding window of size n. Our is deterministic and does not make any assumption on the distribution item frequencies. It supports O(1/ε) update query time, uses space. very simple; its main data structures are just few short queues whose entries store position some in window. We also extend our variable-size This extended O(1/ε log(εn))
Top-N personalized recommendation has been extensively studied in assisting learners finding interesting courses MOOCs. Although existing methods have achieved comparable performance, these models two major shortcomings. First, seldom learn an explicit representation of the structural relation items. Second, most typically obtain a user's general preference and neglect recency This paper proposes Recommendation with Graph Neural Network (TP-GNN) Massive Open Online Course (MOOCs) as solution...
This paper extends the study of online algorithms for energy-efficient deadline scheduling to overloaded setting. Specifically, we consider a processor that can vary its speed between 0 and maximum T minimize energy usage (of which rate is roughly cubic function speed). As upper bounded, system may be with jobs no meet deadlines all jobs. An optimal schedule expected maximize throughput, furthermore, should smallest among schedules achieve throughput. In designing algorithm, one has face...
Allowing students to ask questions in a university course is crucial aspect of learning, which leads increased learning effectiveness but also workload the teaching staff. To reduce their workload, this paper presents design chatbot for instantly answering students' on multiple common social platforms including Telegram, Facebook Messenger and Line. The can answer natural language commands. Once teachers upload necessary course-related information an online database, materials logistics...
Recently, graph neural networks (GNNs) have achieved promising results in session-based recommendation. Existing methods typically construct a local session and global to explore complex item transition patterns. However, studies seldom investigated the repeat consumption phenomenon graph. In addition, it is challenging retrieve relevant adjacent nodes from whole training set owing computational complexity space constraints. this study, we use GNN jointly model intra- inter-session...
Augmented Reality (AR) is a technology that augments the real physical world with computer-generated 3D virtual objects such users can interact them using screen of their mobile devices. This paper studies how to effectively use AR enhance learning experience kindergarten students, while addressing parents' concern long-time usage electronic devices may affect child's health. We developed an application prototype teach students English vocabulary in interactive and attractive way. It allows...
This article extends the study of online algorithms for energy-efficient deadline scheduling to overloaded setting. Specifically, we consider a processor that can vary its speed between 0 and maximum T minimize energy usage (the rate is believed be cubic function speed). As upper bounded, may with jobs no guarantee meet deadlines all jobs. An optimal schedule expected maximize throughput, furthermore, should smallest among schedules achieve throughput. In designing algorithm, one has face...
Energy usage has been an important concern in recent research on online scheduling. In this paper we extend the study of tradeoff between flow time and energy from single-processor setting [8, 6] to multi-processor setting. Our main result is analysis a simple non-migratory algorithm called CRR (classified round robin) m ≥ 2 processors, showing that its plus within O(1) times optimal offline algorithm, when maximum allowable speed slightly relaxed. This still holds even if comparison made...
To tackle the exponentially increasing throughput of Next-Generation Sequencing (NGS), most existing short-read aligners can be configured to favor speed in trade accuracy and sensitivity.SOAP3-dp, through leveraging computational power both CPU GPU with optimized algorithms, delivers high sensitivity simultaneously.Compared widely adopted including BWA, Bowtie2, SeqAlto, CUSHAW2, GEM GPU-based BarraCUDA CUSHAW, SOAP3-dp was found two tens times faster, while maintaining highest lowest false...
We study online nonclairvoyant speed scaling to minimize total flow time plus energy. first consider the traditional model where power function is $P(s)=s^\alpha$. give a algorithm that shown be $O(\alpha^3)$-competitive. then show an $\Omega( \alpha^{1/3-\epsilon} )$ lower bound on competitive ratio of any algorithm. also there are functions for which no can $O(1)$-competitive.
This paper explores the possibility of using augmented reality (AR) technology to improve learning experience kindergarten students, teaching teachers and address parents' concern that a long-time usage electronic devices may affect their child's health. We developed two AR mobile application prototypes teach students English vocabulary in an interactive attractive way. The first allows learn any place at time device. To on health, monitoring system is implemented allow parents monitor stop...
Personalized learning has become an important and powerful paradigm catering for various needs, styles, preferences, modes of learning. Several methods including task recommendations path planning have recently emerged to effectively implement personalized using e-learning systems. The literature shows that the use in systems is a very effective way facilitate vocabulary One key research issues regarding these how model logs each learner. Specifically, measure effectiveness learned tasks...