A5100459588- Topic Modeling
- Face and Expression Recognition
- Advanced Clustering Algorithms Research
- Rough Sets and Fuzzy Logic
- Natural Language Processing Techniques
- Recommender Systems and Techniques
- Advanced Graph Neural Networks
- Advanced Computing and Algorithms
- Educational Technology and Pedagogy
- Geoscience and Mining Technology
- Advanced Neural Network Applications
- Chaos-based Image/Signal Encryption
- Advanced Computational Techniques and Applications
- Multimodal Machine Learning Applications
- Advanced Image and Video Retrieval Techniques
- Adversarial Robustness in Machine Learning
- Anomaly Detection Techniques and Applications
- Privacy-Preserving Technologies in Data
- Employer Branding and e-HRM
- Redox biology and oxidative stress
- Organizational Management and Leadership
- Machine Learning and ELM
- Teleoperation and Haptic Systems
- Mobile Crowdsensing and Crowdsourcing
- Business Process Modeling and Analysis
University of California, Davis
2024
Henan University of Science and Technology
2009-2023
Yancheng Institute of Technology
2023
Xi'an Jiaotong University
2021-2023
Baidu (China)
2023
Harbin Engineering University
2023
Beijing Jiaotong University
2017-2023
Center for Life Sciences
2022
Shanghai Institute of Nutrition and Health
2022
Peking University
2022
Building an effective adversarial attacker and elaborating on countermeasures for attacks natural language processing (NLP) have attracted a lot of research in recent years. However, most the existing approaches focus classification problems. In this paper, we investigate defenses structured prediction tasks NLP. Besides difficulty perturbing discrete words sentence fluency problem faced by attackers any NLP tasks, there is specific challenge to models: output models sensitive small...
The emerging intelligent transportation systems puts higher demands on the collection and analysis of traffic data. LiDAR can provide high-precision point clouds objects, making it a promising choice for surveillance device. This article focuses object detection with roadside LiDAR: estimating both positions categories them. To overcome challenges posed by clouds, we propose GC-net, which is based three-stage pipeline, including gridding, clustering, classification. First, design one-to-one...
Large language models (LLMs) have demonstrated exceptional performance in various natural processing tasks, yet their efficacy more challenging and domain-specific tasks remains largely unexplored. This paper presents FinEval, a benchmark specifically designed for the financial domain knowledge LLMs. FinEval is collection of high-quality multiple-choice questions covering Finance, Economy, Accounting, Certificate. It includes 4,661 spanning 34 different academic subjects. To ensure...
Along with the expansion and in-depth of application domain cluster analysis, one kind new algorithm called Spectral Clustering has been aroused great concern by scholars, is newly developing technique in field machine learning recent years. Unlike traditional clustering algorithms, this can solve non-convex sphere sample spaces globally optimal solution. This paper introduces principle, induction summary to current research situation as well various domains. Firstly, analysis some...
This study aims to solve the problems of sparse reward, single policy, and poor environmental adaptability in local motion planning task autonomous underwater vehicles (AUVs). We propose a two-layer deep deterministic policy gradient algorithm-based end-to-end perception–planning–execution method overcome challenges associated with training learning approaches that directly output control forces. In this approach, state set is established based on environment information, action...
Diabetes mellitus is a kind of metabolic disease characterized by hyperglycemia resulting from insulin insufficiency and resistance. It has become one the major diseases threatening human health. In this paper, we analyze current R&D status diabetes aspects papers, patents, drugs industrial development. The results show that scientific outcomes are increasing steadily hot topics diabetic complications epidemiological research. terms technology development, large pharmaceutical companies,...
Every major milestone in the history of science, technology and interaction design has stemmed from collision human nature (Katona, 2021). From original paper tape punching, it evolved to keyboard input, now touch operation, voice control, recognition, advanced gesture motion capture, eye recognition technologies, so on. In future, brain-computer interfaces mind will also be implemented. Each technological innovation product upgrade bring about changes way human-machine takes place. terms...
Abstract Clustering is the discovery of latent group structure in data and a fundamental problem artificial intelligence, vital procedure data-driven scientific research over all disciplines. Yet, existing methods have various limitations, especially weak cognitive interpretability poor computational scalability, when it comes to clustering massive datasets that are increasingly available domains. Here, by simulating multi-scale observation process humans, we design scalable algorithm detect...
Learning programming challenges students who encounter difficulties such as resolving syntax and format errors. These require to invoke resilience overcome problems keep trying. In response, this study developed the scale for university (PRSUS). The snowball-sampling method was used collect data of science engineering undergraduates, participants were divided into two groups (n1 = 316, n2 358) an exploratory a confirmatory factor analysis, respectively. For PRSUS, 3 items retained each 4...
Cloud segmentation is a fundamental step in accurately acquiring cloud cover. However, due to the nonrigid structures of clouds, traditional methods perform worse than expected. In this paper, novel deep convolutional neural network (CNN) named MA-SegCloud proposed for segmenting images based on multibranch asymmetric convolution module (MACM) and an attention mechanism. The MACM composed convolution, depth-separable squeeze-and-excitation (SEM). not only enables capture more contextual...
BACKGROUND:Osteosarcoma (OS) is very common worldwide, and the mechanisms underlying its development remain unclear. This study aims to identify key genes promoting reproduction, invasion, transfer of osteosarcoma cells. MATERIAL AND METHODS:Gene expression profile data (GSE42352 GSE42572) were downloaded from Gene Expression Omnibus database. Differentially expressed calculated using R software. ontology enriched pathway analysis mRNAs analyzed by FunRich. Verification was conducted...
Similarity between objects is multi-faceted and it can be easier for human annotators to measure when the focus on a specific aspect. We consider problem of mapping into view-specific embeddings where distance them consistent with similarity comparisons form "from t-th view, object A more similar B than C". Our framework jointly learns exploiting correlations views. Experiments number datasets, including one multi-view crowdsourced comparison bird images, show proposed method achieves lower...
Abstract. This paper explores the role of cryptography and encryption technology in evolution digital rights management (DRM) gaming industry. A collection guidelines, procedures, instruments that control appropriate use content is collectively referred to as (DRM). Cryptography plays a vital DRM by ensuring data privacy, authenticity, integrity. The early stages involved physical disc-based methods, such unique disc features online activation. However, these methods faced challenges...
The paper presents combining k-error approximate entropy with some tests in NIST's STS randomness test suite to analyze cryptographical properties of several chaotic pseudorandom sequences generated by diverse maps and quantified methods, including Logistic map, Cubic Henon Sine Tent Chebyshev Piecewise-linear Piecewise-square-root map. Simulation results show that the can distinguish stability different sequences; cryptological are influenced methods; map have best properties, while worst stability.
Federated Learning (FL) has attracted much attention in recent years as a promising framework of privacy-preserving and scalability. However, the is highly vulnerable to data heterogeneity, which widely exists real-world applications. In this paper, we designed communication-efficient personalized federated clustering algorithm k-PFed, overcomes heterogeneity various network scenarios FL system. By initializing local datasets by pre-clustering, clients k-PFed are able generate model locally....