A5100736536- Advanced Computational Techniques and Applications
- Brain Tumor Detection and Classification
- Natural Language Processing Techniques
- Advanced Neural Network Applications
- Advanced Bandit Algorithms Research
- Artificial Intelligence in Healthcare
- Topic Modeling
- Distributed and Parallel Computing Systems
- Neonatal and fetal brain pathology
- Advanced Image and Video Retrieval Techniques
- Advanced Technologies in Various Fields
- Recommender Systems and Techniques
- Digital Media Forensic Detection
- Image Enhancement Techniques
- Anomaly Detection Techniques and Applications
- Misinformation and Its Impacts
- Infrastructure Maintenance and Monitoring
- Advanced Database Systems and Queries
- Machine Learning in Healthcare
- Artificial Intelligence in Games
- Remote Sensing and LiDAR Applications
- Service-Oriented Architecture and Web Services
- Data Quality and Management
- Geophysical Methods and Applications
- Sentiment Analysis and Opinion Mining
Renmin University of China
2021-2024
Power Grid Corporation (India)
2024
Guilin University of Electronic Technology
2024
Tencent (China)
2023
Hebei University
2011-2023
China South Industries Group (China)
2021
Southwest Jiaotong University
2021
Institute of High Energy Physics
2021
Capital Medical University
2020
Beijing YouAn Hospital
2020
In the context of neuro-pathological disorders, neuroimaging has been widely accepted as a clinical tool for diagnosing patients with Alzheimer’s disease (AD) and mild cognitive impairment (MCI). The advanced deep learning method, novel brain imaging technique, was applied in this study to evaluate its contribution improving diagnostic accuracy AD. Three-dimensional convolutional neural networks (3D-CNNs) were magnetic resonance (MRI) execute binary ternary classification models. dataset...
Through well-designed counterfeit websites, phishing induces online users to visit forged web pages obtain their private sensitive information, e.g., account number and password. Existing antiphishing approaches are mostly based on page-related features, which require crawl content of as well accessing third-party search engines or DNS services. This not only leads low efficiency in detecting but also makes them rely network environment services heavily. In this paper, we propose a fast...
The popularity of digital images is rapidly increasing due to improving imaging technologies and convenient availability facilitated by the Internet. However, how find user-intended from Internet nontrivial. main reason that Web are usually not annotated using semantic descriptors. In this article, we present an effective approach a prototype system for image retrieval mining. can also serve as search engine. One key ideas in extract text information on pages semantically describe images....
Multi-label few-shot image classification is a crucial and challenging task due to limited annotated data elusive category specificity. However, research on this topic still in the rudimentary stage few methods are available. Existing either leverage augmentation alleviate scarcity or utilize label features as auxiliary knowledge eliminate negative effect caused by irrelevant categories, but they ignore utilization of region for augmentation, overlook learn appropriate text feature better...
To establish an early diagnostic system for hypoxic ischemic encephalopathy (HIE) in newborns based on artificial neural networks and to determine its feasibility. Based published research as well preliminary studies our laboratory, multiple noninvasive indicators with high sensitivity specificity were selected the diagnosis of HIE employed present study, which incorporates fuzzy logic networks. The analysis results from network experiments 140 cases showed a correct recognition rate 100%...
Convolutional neural networks (CNNs), as representatives of deep learning, are one the most commonly used in applications such graphic image analysis. However, CNN has heavy computation patterns; network training processes could take several hours even with modern processors. Different from process, inference process is more often executed on devices low computing power, CPUs. Fortunately, a minimal filtering algorithm, Winograd, can reduce convolution computations by reducing number...
Ms. Pac-Man is a popular chasing and evading game the ghost character in controlled by script. This article evolved an evolutionary neural network for red to chase Pac-Man. Red ghost' position, Pac-Man's position state are considered be inputs of network, output direction move next step. We also proposed fitness function raise capture ability evolution so that learns itself simulation. Experimental results show agent well plays better teamwork than traditional script ghost.
Road markings are an important guarantee for road infrastructure and driving safety. In recent years, the rapid development of autonomous technology, identification ofroad identifiers has become indispensable part this technology. The traditional target detection algorithm in computer vision poor regional selectivity, high time complexity feature robustness. deep learning can effectively solve these problems. At present, is widely used audio data images, it feasible meaningful to identify...
For click-through rate (CTR) prediction tasks, a good performance can be obtained by full explorations of both user behavior and item behavior. Since user’s interests have great influence on behaviors, it is very important to learn users’ intrinsic according their behaviors. User are not only diverse but also in dynamic change. However, the dynamics interests’ change fully taken into account majority current CTR models. The latest sequential recommendation algorithm ignores subjectivity...
With the development of artificial intelligence technology, intelligent weapon systems that can automatically identify, lock on and strike targets have gradually appeared replace humans in executing simple decision-making commands. Target detection is a key part weapons. At present, large-scale target has serious challenges such as long-tail data distributions, severe occlusion, category ambiguity. The main algorithms only detect each independent area without considering semantic...
Abstract The cores of granular computing are the granule, granule layer and structure. In this paper concepts “Ontology granule” “Compatible were defined, applying ideas an ontology model, Ontology set; thus tree generation algorithms proposed. These produce initial set with a compatible class, extend other granules by vector intension IG, building lattice hierarchy conception model relation RG. empirical research traditional Chinese medicine shows that these correct efficient, provide good...
As a free online encyclopedia with large-scale of knowledge coverage, rich semantic information and quick update speed, Wikipedia brings new ideas to measure correlation. In this paper, we present method for measuring the correlation between words by mining that exists in Wikipedia. Unlike previous methods calculate relatedness merely based on page network or category network, our not only takes into account also combines it improve accuracy results. Besides, analyze evaluate algorithm...
Data science pipelines commonly utilize dataframe and array operations for tasks such as data preprocessing, analysis, machine learning. The most popular tools these are pandas NumPy. However, limited to executing on a single node, making them unsuitable processing large-scale data. Several systems have attempted distribute applications clusters while maintaining interfaces similar single-node libraries, enabling scientists scale their workloads without significant effort. existing often...
Identification of proton and gamma plays an essential role in ultra-high energy gamma-ray astronomy with LHAASO-KM2A. In this work, two neural networks (deep (DNN) graph (GNN)) are applied to distinguish the LHAASO-KM2A simulation data. The receiver operating characteristic (ROC) curves used evaluate quality model. Both KM2A-DNN KM2A-GNN models give higher Area Under Curve (AUC) scores than traditional baseline
In order to determine the work space of a small reverberation chamber, experimental system is established measure. Electric field intensity in eight positions test measured and standard deviations electric spaces calculated, result shows that satisfies demand IEC 61000-4-21.