A5100423230- Recommender Systems and Techniques
- Advanced Graph Neural Networks
- Data Stream Mining Techniques
- Online Learning and Analytics
- Expert finding and Q&A systems
- Helicobacter pylori-related gastroenterology studies
- Data Mining Algorithms and Applications
- Speech and Audio Processing
- Human Mobility and Location-Based Analysis
- Web Data Mining and Analysis
- Data Management and Algorithms
- Digital Media and Visual Art
- Innovative Teaching Methods
- Embedded Systems Design Techniques
- Caching and Content Delivery
- Advanced Neural Network Applications
- Bioinformatics and Genomic Networks
- Advanced Decision-Making Techniques
- Radiology practices and education
- Metastasis and carcinoma case studies
- Web Application Security Vulnerabilities
- Stock Market Forecasting Methods
- Protein Structure and Dynamics
- Context-Aware Activity Recognition Systems
- Time Series Analysis and Forecasting
China Internet Network Information Center
2020-2023
University of Manchester
2022
Nanjing Normal University
2021
Shanghai Electric (China)
2020
Wuhan Institute of Technology
2020
Zhengzhou University of Light Industry
2017
Nanyang Technological University
2014
Fudan University
2010
PLA Army Engineering University
2010
Deep learning-based recommender systems (DLRSs) often have embedding layers, which are utilized to lessen the dimension of categorical variables (e.g., user/item identifiers) and meaningfully transform them in low-dimensional space. The majority existing DLRSs empirically pre-define a fixed unified for all embeddings. It is evident from recent researches that different sizes highly desired users/items according their frequency. However, manually selecting can be very challenging due large...
Cross-site scripting (XSS) is one of the main threats Web applications, which has great harm. How to effectively detect and defend against XSS attacks become more important. Due malicious obfuscation attack codes gradual increase in number, traditional detection methods have some defects such as poor recognition codes, inadequate feature extraction low efficiency. Therefore, we present a novel approach based on attention mechanism Long Short-Term Memory (LSTM) recurrent neural network. First...
Deep learning based recommender systems (DLRSs) often have embedding layers, which are utilized to lessen the dimensionality of categorical variables (e.g. user/item identifiers) and meaningfully transform them in low-dimensional space. The majority existing DLRSs empirically pre-define a fixed unified dimension for all embeddings. It is evident from recent researches that different sizes highly desired users/items according their popularity. However, manually selecting can be very...
Recommendation system can greatly alleviate the “information overload” in big data era. Existing recommendation methods, however, typically focus on predicting missing rating values via analyzing user-item dualistic relationship, which neglect an important fact that latent interests of users influence their behaviors. Moreover, traditional methods easily suffer from high dimensional problem and cold-start problem. To address these challenges, this paper, we propose a PBUED (PLSA-Based...
Abstract Background To improve the eradication rate of H. pylori , researchers have investigated role WeChat-based mini-app as an electronic reminding system in treatment. Methods Subjects from three medical centers were divided into two groups. Patients daily mini-app-based notification group received notifications via WeChat mini-app. control one-time verbal education on first clinical visit. Both groups a 14-day quadruple therapy to eradicate infection. Eradication rate, compliance,...
User Navigation Behavior Mining (UNBM) mainly studies the problems of extracting interesting user access patterns from sequences (UAS), which are usually used for prediction and web page recommendation. Through analyzing real world data, we find most carrying hybrid features different patterns, rather than a single one. Therefore, methods that categorize one sequence into pattern, can hardly obtain good quality results. Due to this problem, propose multi-task learning approach based on...
Extracting entities and their relationships from financial documents is crucial for analyzing predicting future market trends. However, the current state of art in this field faces two major challenges: multiple sentences between related poor few-shot performance caused by vast amount knowledge required domain. To address these challenges, we propose a framework entity-relation extraction that leverages multi-turn machine reading comprehension (MRC) Longformer model to handle long text...
The frequent items problem is to process a stream as of and find all occurring more than given fraction the time. It one most heavily studied problems in data mining, dating back 1980s. Aiming at higher false positive rate Space-Saving algorithm, an LRU-based (Least Recently Used, LRU) improved algorithm with low frequency item pre-eliminated proposed. Accuracy, stability adaptability have been apparently enhanced. Experimental results indicate that can not only be used items, estimate them...
Background: A pink color change occasionally found by us under magnifying endoscopy with narrow-band imaging (ME-NBI) may be a special feature of early gastric cancer (EGC), and was designated the "pink pattern". The purposes this study were to determine relationship between pattern cytopathological changes in cells whether is useful for diagnosis EGC. Methods: features ME-NBI images pathological cancerous mucosal surfaces extracted quantified. cosine similarity calculated evaluate...
Abstract Background: Because it is so laborious and expensive to experimentally identify Drug-Target Interactions (DTIs), only a few DTIs have been verified. Computational methods are useful for identifying in biological studies of drug discovery development. Results: For drug-target interaction prediction, we propose novel neural network architecture, DAEi, extended from Denoising AutoEncoder (DAE). We assume that set verified corrupted version the full set. use DAEi learn latent features...
MOOC is becoming increasingly popular around the world. However, its status and effectiveness in higher education are still controversial. This study investigates Chinese postgraduate students' learning experiences (factors driving them to learn future optimisation suggestions) on University (iCourse) platform. By conducting semi-structured interviews with five graduate students, data was gathered. According findings, students choose because of flexibility their need for professional...
MOOCs, as a new educational mode and form, has been widely welcomed around the world also flourished in China for nearly ten years. Under background that largest total amount of MOOCs world, it is great significance to explore characteristics high-quality presentation future development. At present, although many studies have proved MOOCs’ teaching video an impact on effect there are few answers questions such “how present” “why present this way”. Based this, from perspective qualitative...
This paper presents the application of Dynamic Time Warping (DTW) algorithm in analysis human functional movements activities daily living (ADLs).Dynamic was originally developed for automatic speech recognition, though method has been adopted by several fi elds biomechanics.As a part post-stroke rehabilitation project COSMOSYS, aim is to quantify ADL performances hemiparetic subjects, hence be able track their progress during physiotherapy.
Deep neural networks (DNNs) that incorporated lifelong sequential modeling (LSM) have brought great success to recommendation systems in various social media platforms. While continuous improvements been made domain-specific LSM, limited work has done cross-domain which considers of sequences both target domain and source domain. In this paper, we propose Lifelong Cross Network (LCN) incorporate LSM improve the click-through rate (CTR) prediction The proposed LCN contains a LifeLong...
Retargeting aims to shrink a photo wherein the perceptually prominent regions are appropriately kept. In practice, optimally shrinking high resolution (HR) aerial is useful tool for smart navigation. Nowadays, vehicle drivers' path planning generally guided by an HR recommended navigation App like Google Maps. Owing limited and various of displays, we have retarget each original accordingly, navigation-aware can be well preserved. retargeting non-trivial due three challenges: 1) rich number...