Shiding Sun

ORCID: 0000-0003-4590-4479
Publications
Citations
Views
---
Saved
---
About
Contact & Profiles
Research Areas
  • Image Retrieval and Classification Techniques
  • Traditional Chinese Medicine Studies
  • Text and Document Classification Technologies
  • Machine Learning and Data Classification
  • Machine Learning in Healthcare
  • Advanced Neural Network Applications
  • Advanced Image and Video Retrieval Techniques
  • Domain Adaptation and Few-Shot Learning
  • Software-Defined Networks and 5G
  • Face and Expression Recognition
  • Machine Learning and ELM
  • Imbalanced Data Classification Techniques
  • Advanced Welding Techniques Analysis
  • Video Surveillance and Tracking Methods
  • Cutaneous Melanoma Detection and Management
  • Traditional Chinese Medicine Analysis
  • Phase Change Materials Research
  • Laser Applications in Dentistry and Medicine
  • Smart Agriculture and AI
  • Digital Imaging for Blood Diseases
  • Artificial Intelligence in Healthcare
  • Surface Treatment and Residual Stress
  • Remote-Sensing Image Classification
  • Advanced Computing and Algorithms
  • Welding Techniques and Residual Stresses

China United Network Communications Group (China)
2023-2024

University of Chinese Academy of Sciences
2021-2024

Beijing Institute of Big Data Research
2022-2024

Chinese Academy of Sciences
2022-2024

The University of Sydney
2019

Renmin University of China
2018-2019

Seoul National University
2019

University of Nebraska–Lincoln
2017-2018

Tsinghua University
2011-2014

Nano and Advanced Materials Institute
2011

Deep convolutional neural networks (DCNNs) can classify skin diseases at a level equivalent to dermatologist, but their performance in specific areas requires further research. To evaluate the of trained DCNN‐based algorithm classifying benign and malignant lip diseases. A training set 1629 images (743 malignant, 886 benign) was used with Inception‐Resnet‐V2. Performance evaluated using another 344 281 from other hospitals. Classifications by 44 participants (six board‐certified...

10.1111/bjd.18459 article EN British Journal of Dermatology 2019-08-26

One way to extract patterns from clinical records is consider each patient record as a bag with various number of instances in the form symptoms. Medical diagnosis discover informative ones first and then map them one or more diseases. In many cases, patients are represented vectors some feature space classifier applied after generate results. However, real-world data often low-quality due variety reasons, such consistency, integrity, completeness, accuracy, etc. this paper, we propose novel...

10.1109/ijcnn.2019.8851846 article EN 2022 International Joint Conference on Neural Networks (IJCNN) 2019-07-01

10.1016/j.neunet.2021.11.002 article EN Neural Networks 2021-11-15

10.1016/j.patcog.2024.110377 article EN Pattern Recognition 2024-03-02

One way to extract patterns from clinical records is consider each patient record as a bag with various number of instances in the form symptoms. Medical diagnosis discover informative ones first and then map them one or more diseases. In many cases, patients are represented vectors some feature space classifier applied after generate results. However, real-world data often low-quality due variety reasons, such consistency, integrity, completeness, accuracy, etc. this paper, we propose novel...

10.48550/arxiv.1904.04460 preprint EN other-oa arXiv (Cornell University) 2019-01-01

Syndrome differentiation in Traditional Chinese Medicine (TCM) is the process of understanding and reasoning body condition, which essential step premise effective treatments. However, due to its complexity lack standardization, it challenging achieve. In this study, we consider each patient's record as a one-dimensional image symptoms pixels, missing negative values are represented by zero labeled one or more syndromes diabetes. The objective find relevant first then map them proper...

10.1109/bibm.2018.8621344 article EN 2021 IEEE International Conference on Bioinformatics and Biomedicine (BIBM) 2018-12-01

We use tensile–shear tests to investigate the failure modes of Ti–1Al–1Mn microscale resistance spot welds and determine how mode affects microstructure, microhardness profile, mechanical performance. Two different were revealed: interfacial pullout mode. The that fail by have much better properties than those results show weld nugget size is also a principal factor determines welds. A minimum exists above which all specimens However, critical sizes calculated using existing recommendations...

10.1177/1687814018785283 article EN cc-by Advances in Mechanical Engineering 2018-07-01

10.1109/wcnc57260.2024.10571005 article EN 2022 IEEE Wireless Communications and Networking Conference (WCNC) 2024-04-21

With the progress of human society and development computer technology, more experience has been accu¬mulated, which can serve as prior information in model training. assistance information, performance some machine learning algorithms significantly improved. In this article, we mainly discussed four types namely knowledge sets, fuzzy universum data, privilege information. Specifically, (1) sets are constraints composed domain experience; (2) provide importance each instance through...

10.1016/j.procs.2022.01.033 article EN Procedia Computer Science 2022-01-01

Syndrome differentiation in Traditional Chinese Medicine (TCM) is the process of understanding and reasoning body condition, which essential step premise effective treatments. However, due to its complexity lack standardization, it challenging achieve. In this study, we consider each patient's record as a one-dimensional image symptoms pixels, missing negative values are represented by zero pixels. The objective find relevant first then map them proper syndromes, that similar object...

10.48550/arxiv.1812.07764 preprint EN other-oa arXiv (Cornell University) 2018-01-01

We study the problem of structured output prediction. Some methods such as support vector machines (SSVM) and conditional random fields (CRFs) are state-of-the-art in dealing with data. However, these classical have some limits scalability because high memory requirements slow training speed. Recently, method joint kernel estimation (JKSE) has been proposed based on one-class SVM which can be trained efficiently. JKSE is not powerful those from point prediction performance. To improve...

10.1109/access.2019.2927693 article EN cc-by IEEE Access 2019-01-01
Coming Soon ...