A5100337759- Software-Defined Networks and 5G
- Coronary Interventions and Diagnostics
- Network Security and Intrusion Detection
- Advanced Computational Techniques and Applications
- Adversarial Robustness in Machine Learning
- Advanced Vision and Imaging
- Advanced MIMO Systems Optimization
- Advanced Database Systems and Queries
- Healthcare Systems and Reforms
- Data Visualization and Analytics
- Topic Modeling
- Global Public Health Policies and Epidemiology
- School Health and Nursing Education
- Influenza Virus Research Studies
- Software Testing and Debugging Techniques
- Time Series Analysis and Forecasting
- Web Applications and Data Management
- Advanced Image Processing Techniques
- Optical measurement and interference techniques
- Optimization and Search Problems
- Advanced Sensor and Control Systems
- Photoacoustic and Ultrasonic Imaging
- Image Processing Techniques and Applications
- Advanced Control Systems Optimization
- Hate Speech and Cyberbullying Detection
University of Technology Malaysia
2024-2025
Southeast University
2024
Stanford University
2022-2024
Shanghai Key Laboratory of Trustworthy Computing
2023
East China Normal University
2022-2023
Jilin University
2007-2022
Capital Normal University
2022
Jilin Province Science and Technology Department
2022
Changzhou Institute of Technology
2022
Waseda University
2022
Skin cancer is one of the most common types globally. Despite remarkable advancements deep learning methods in computer vision, automatic diagnosis skin diseases still faces challenges such as limited data and class imbalance. Generative Adversarial Networks (GANs), which can synthesize realistic data, appear an alternative to mitigate these issues. However, for imbalanced unconditional GANs either generate uneven distribution or neglect universal knowledge whole dataset, while...
Importance Influenza vaccination rates remain low among primary school students and vary by in Beijing, China. Theory-informed, multifaceted strategies are needed to improve influenza uptake. Objective To evaluate the effectiveness of improving uptake students. Design, Setting, Participants This cluster randomized trial was conducted from September 2022 May 2023 across schools Schools were allocated randomly a 1:1 ratio or usual practice. deemed eligible if 2019 2020 season fell at below...
Alignment tuning is crucial for ensuring large language models (LLMs) behave ethically and helpfully. Current alignment approaches require high-quality annotations significant training resources. This paper proposes a low-cost, tuning-free method using in-context learning (ICL) to enhance LLM alignment. Through an analysis of ICL demos, we identified style as key factor influencing capabilities explicitly restyled exemplars based on this stylistic framework. Additionally, combined the demos...
The wide field of view fisheye cameras introduces significant image distortion, making accurate depth estimation more challenging compared to pinhole camera models. This paper proposes a panoramic network based on edge awareness, aimed at improving accuracy for images. We design an Edge-Aware Module (EAM) that dynamically weights features extracted by Residual Convolutional Neural Network (Residual CNN) using the information. Subsequently, spherical alignment method is used map from...
Nowadays, with the appearance of more and web services, it has been one key points that how to find target service quickly precisely. Traditional methods discovery are only based on keyword matching, but it's very difficult realize detailed intelligent results hardly meet user's demands. According current OWL-S, WSMO other frameworks, a new lightweight semantic extension framework for services is proposed in this paper, which founded syntax structure XML 3-layer frame can be compatible UDDI....
Generic `toxicity' classifiers continue to be used for evaluating the potential harm in natural language generation, despite mounting evidence of their shortcomings. We consider challenge measuring misogyny and argue that generic are inadequate this task. use data from two well-characterised `Incel' communities on Reddit differ primarily degrees construct a pair training corpora which we fine-tune models. show an open source classifier is unable distinguish meaningfully between generations...
School shutdown due to the global pandemic of coronavirus disease 2019 (COVID-19) can lead an increase in education inequality through disproportionately affecting disadvantaged children. We use data from a unique survey 7,330 junior high school students and their parents Shaanxi explore parental socioeconomic status that enlarges educational gap during pandemic. Comparing students’ test scores before after by characteristics, we find rather than income causes wider gap. One year decrease...
In the context of rapid development sharing economy, shared bikes as its representative products rise rapidly. Among many problems, problem bicycle stock scheduling is most severe, which will directly bring great troubles to People's Daily life. This paper compares and understands various demand forecasting methods by minimizing cost objective function. By using BP neural network, this provides reference for following data bike system, builds system solve problem.
The capacity and the energy consumption of last-level cache (LLC) must be improved toward future microprocessors. This paper proposes an STT-RAM-based hybrid architecture (HCA) its replacement policy, named Fast Region First Used (FRFU) policy. proposed HCA is composed two types STT-RAM devices with different characteristics to mitigate disadvantages non-volatile devices. FRFU policy provides better handling data blocks for than LRU evaluation results show that can successfully reduce by 55%...
For the widely existing document and relation mixed-type data, this paper proposes a system of storage query for associated data based on distributed middleware. It adopts middleware as technological tools to integrate databases different types, which achieves adaptive efficient performance type data. Furthermore, hybrid mechanism is provided extend object SQL statement. In addition, extended operators such fuzzy join are implemented realize between proposed system, upper layer uses...
This paper constructs a dual-sparse optimal mean-variance factor portfolio to improve the interpretation of in set 187 anomaly portfolios. The method limits nonzero elements principal components (PCs) and number invested cross-section. We first impose substantive sparsity on PCs with only around 20 elements, or 10% underlying factors, high degree facilitates interpret their economic meaning at little cost variance explained. then shrink cross-section most 6 sparse can almost approximate (MV)...
Abstract To predict the deep favorable gas zone in Songliao Basin during fault period is an important component of work on expanding prospect area Basin. In order to carry out large‐scale oil and exploration Basin, we chose Shengping‐Xingcheng structure within Xujiaweizi depression as resource base perform ground three‐component seismic exploration, reached a feasible prediction method . Using geologic data with control exploratory wells, data, P‐wave P‐SV wave layer calibration, phase...
This paper analyzes the current situation of non-contact measurement at home and abroad, a profound understanding importance data synchronous measuring machine.Then, design three types synchronization experiment, set up verification environments respectively to analyze character in systems.Finally, sending pulse by hardware can realize coordinate machine.
Abstract Similarity‐based exploration is an effective method in knowledge discovery. Faced with multivariate event sequence data (MVES), developing a satisfactory similarity measurement for specific question challenging because of the heterogeneity introduced by numerous attributes different formats, coupled their associations. Additionally, absence validation feedback makes judging goodness scheme time‐consuming and error‐prone procedure. To free analysts from tedious programming to...
Resource allocation of 5G cellular network slicing has been gaining huge interest recently. However, it is not well-explored under the Multi-tenant and Multi-service Coexistence Network scenario. This paper proposes a dynamic scheme for Network, which considers tenants, priority, baseband resources, slice granularity, service characteristics, quality (QoS), fairness index. The framework consists group-level, performs clustering preprocessing, setting priority service, filtering desired...
Adversarial training is an efficacious defense approach to protect classification model against adversarial attacks. In this paper, we reveal that a significant difference exists between the feature map of original sample and its corresponding version. Based on main insight, propose novel robust feature-based examples called FPAT, where are generated by maximizing loss function clean maps. We show via extensive experiments MNIST, SVHN CIFAR-10, our proposed method as effective...