A5027072710- Topic Modeling
- Natural Language Processing Techniques
- Multimodal Machine Learning Applications
- Domain Adaptation and Few-Shot Learning
- Web Data Mining and Analysis
- Advanced Text Analysis Techniques
- Network Security and Intrusion Detection
- Metaheuristic Optimization Algorithms Research
- Corporate Finance and Governance
- Complex Network Analysis Techniques
- Financial Markets and Investment Strategies
- Speech and dialogue systems
- Educational Technology and Pedagogy
- Anomaly Detection Techniques and Applications
- Advanced Graph Neural Networks
- Machine Learning in Healthcare
- Text and Document Classification Technologies
- Image and Object Detection Techniques
- Advanced Image and Video Retrieval Techniques
- Advanced Algorithms and Applications
- Auditing, Earnings Management, Governance
- Advanced Computational Techniques and Applications
- Sentiment Analysis and Opinion Mining
- Spam and Phishing Detection
- Machine Learning and ELM
Zhejiang University
2009-2025
Yanbian University
2025
Chinese University of Hong Kong, Shenzhen
2025
Henan University of Science and Technology
2024
Southern University of Science and Technology
2024
Leshan Normal University
2011-2023
Agricultural Bank of China
2022
Huawei Technologies (Sweden)
2020-2021
Shenyang Jianzhu University
2021
Chengdu Normal University
2014-2020
Recommender systems have shown great potential to solve the information explosion problem and enhance user experience in various online applications. To tackle data sparsity cold start problems recommender systems, researchers propose knowledge graphs (KGs) based recommendations by leveraging valuable external as auxiliary information. However, most of these works ignore variety types (e.g., texts images) multi-modal (MMKGs). In this paper, we Multi-modal Knowledge Graph Attention Network...
While deep learning demonstrates its strong ability to handle independent and identically distributed (IID) data, it often suffers from out-of-distribution (OoD) generalization, where the test data come another distribution (w.r.t. training one). Designing a general OoD generalization framework for wide range of applications is challenging, mainly due different kinds shifts in real world, such as shift across domains or extrapolation correlation. Most previous approaches can only solve one...
Rui Sun, Yue Zhang, Meishan Donghong Ji. Proceedings of the 53rd Annual Meeting Association for Computational Linguistics and 7th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2015.
Generative models in AI are an entirely new paradigm for machine learning, allowing computers to create realistic data all kinds of categories, like text (NLP), images, and even physics simulations. In this paper formalism is used guide the theory, algorithms applications generative models, with particular focus on a few well established techniques VAEs, GANs, diffusion models. It stresses importance probabilistic modelling information theory (I.e. KL divergence, ELBO, adversarial...
Abstract In modern warfare, increasingly complex terrain and road conditions pose great challenges to the actual passing speed maneuverability performance of combat vehicles, especially in terms soft-surface performance. this paper, for single condition influencing factors existing study maneuverability, based on joint simulation method multi-body dynamics-discrete element (MBD-DEM), we analyze effect tires, front rear tire traction performance, their differences under a different relative...
Plant defense against herbivores is primarily regulated by the phytohormone jasmonate (JA). At core, JA signaling MYC2 transcription factor (TF) that regulates expression of an extensive array defense-related genes. However, regulatory mechanisms underlying MYC2-mediated herbivore resistance in rice are not fully understood. We employed brown planthopper (BPH) bioassays, transcriptional activation assays, transcriptome profiling, targeted metabolomics and cleavage under targets...
Using genotype data consisting of 8,316 individuals, we systematically evaluated imputation performance across six state-of-the-art reference panels for Chinese and Thai populations. A substantial proportion variants identified through whole-genome sequencing, especially low-frequency variants, remained undetected by existing panels. In the population, TOPMed panel required an R2 threshold 0.60-0.70 to achieve comparable accuracy ChinaMAP without filtering, challenging standard practice...
<title>Abstract</title> Using data consisting of 8,316 individuals, we evaluated performance genotype imputation across six state-of-the-art reference panels for Chinese and Thai populations. A substantial proportion variants identified through whole-genome sequencing, especially low-frequency variants, remained undetected by these panels. In samples, the TOPMed panel required an R2 threshold 0.60-0.70 to achieve comparable accuracy ChinaMAP without filtering, challenging standard practice...
As one of the most popular social media platforms in China, Weibo has aggregated huge numbers texts containing people's thoughts, feelings, and experiences. Analyzing emotions expressed on attracted a great deal academic attention. Emotion lexicon is vital foundation sentiment analysis, but existing lexicons still have defects such as limited variety emotions, poor cross-scenario adaptability, confusing written online expressions words. By combining grounded theory semi-automatic methods, we...
The number of IoT devices continues to increase, but the security cannot be guaranteed. Many are infected with malware, forming huge botnets, which could launch DDoS attacks and cause heavy losses. In recent years, malware family has a tendency centralized on ARM-based devices. most widely spread families Mirai Gafgyt family. this paper, we automatically extract instruction sequences these two families' samples use as language describe samples. We transfer word vector space by Word2Vec. Then...
合成孔径雷达(SAR)成像处理的运算量较大,在基于中央处理器(Central Processing Unit, CPU)的工作站或服务器上一般需要耗费较长的时间,无法满足实时性要求。借助于通用并行计算架构(CUDA)编程架构,该文提出一种基于图形处理器(GPU)的SAR 成像处理算法实现方案。该方案解决了GPU 显存不足以容纳一景SAR 数据时数据处理环节与内存/显存间数据传输环节的并行化问题,并能够支持多GPU 设备的并行处理,充分利用了GPU设备的计算资源。在NVIDIA K20C 和INTEL E5645 上的测试表明,与传统基于GPU 的SAR 成像处理算法相比,该方案能够达到数十倍的速度提升,显著降低了处理设备的功耗,提高了处理设备的便携性,能够达到每秒约36兆采样点的实时处理速度。
Particle swarm optimization (PSO) is a heuristic stochastic evolutionary algorithm. However, standard PSO exists unbalanced exploitation and exploration, lower convergence speed. An improved technique introduced into the with adaptive computation of inertia weights. After every iteration, new competition random operated to jump out local optimum. Four benchmark functions are selected test validate constructed The numerical experiments results show that proposed algorithm effective. speed...
A crowd behavior identification method is proposed by combining the streakline based on fluid mechanics with a high-accurate variational optical flow model in this paper. Firstly, calculated model, streaklines are used to acquire motion trajectory information. The angular histogram and regions of interest scenes obtained calculating clustering dasymetric dot maps starting ending points trajectory, then, map information analyze whether there specific behaviors interest, thus identify...
Vector data compression algorithm can meet requirements of different levels and scales by reducing the amount vector graphics, so as to reduce transmission, processing time storage overhead data.In view fact that large threshold leading comparatively error in Douglas-Peucker algorithm, which has difficulty maintaining uncertainty shape features selection, a segmented based on node importance is proposed.Firstly, uses vertical chord ratio main feature detect extract critical points with...
With the continuous development of modern information technology, various new technologies are gradually emerging. As an important content discipline computer science, artificial intelligence can simulate human activities through intelligent systems or machines, so as to extend people’s intelligence. Artificial has been widely used in all walks life, combining and marketing effectively innovate traditional methods, improve overall quality marketing, analysis field application status, clear...
The phenomenon of corporate debt default has broken out in recent years, which highlights the importance macro situation to stability business operations. In this paper, A-share listed companies Shanghai and Shenzhen Stock Exchange from 2008 2018 are used as a sample conduct research perspective independent directors with vision. empirical results show that larger number higher proportion macro-background bond issuing companies, less will be found; moreover, relationships stronger when...
The key of the out-of-distribution (OOD) generalization is to generalize invariance from training domains target domains. variance risk extrapolation (V-REx) a practical OOD method, which depends on domain-level regularization but lacks theoretical verifications about its motivation and utility. This article provides insights into V-REx by studying variance-based regularizer. We propose Risk Variance Penalization (RVP), slightly changes addresses theory concerns V-REx. provide explanations...
Based on many cases of publicity using social media in the market, this paper empirically examines damage microblog to market efficiency by sample IPO companies. The results show that companies posting microblogs during period have higher underpricing, and more number microblogs, richer content, proportion operating activities, underpricing. These relationships are stronger enterprises with degree information asymmetry management valuing more. Furthermore, posted mainly influence investor...
Recent studies show structured atomic event information is beneficial to represent the discourse semantic. However, extracting useful representation of events from open domain a challenging problem. On one hand, previous extraction methods on special domain, cannot be directly used for because limitation and predefined pattern. other simply regarded as preprocessing step in related work, few focus domain. In this paper, we propose an unsupervised method Chinese Being directed against...
In engineering practices, curve fitting is a common method to evaluate the performance of machines or equipments using collected data. If data works as whole and can not be divided into groups, conventional methods used perform task. Researchers have proposed some improved with higher precision computational efficiency, in order handle special situations. Another situation, however, involves that consist more than one sample set, which show obvious differences collection methods, districts...