Search

Your search keyword '"Wang, Guiqin"' showing total 528 results
Start Over Author "Wang, Guiqin"
528 results on '"Wang, Guiqin"'

1. EdgeSync: Faster Edge-model Updating via Adaptive Continuous Learning for Video Data Drift

Author

Zhao, Peng, Dong, Runchu, Wang, Guiqin, and Zhao, Cong

Subjects
Computer Science - Computer Vision and Pattern Recognition, Computer Science - Artificial Intelligence
Abstract

Real-time video analytics systems typically place models with fewer weights on edge devices to reduce latency. The distribution of video content features may change over time for various reasons (i.e. light and weather change) , leading to accuracy degradation of existing models, to solve this problem, recent work proposes a framework that uses a remote server to continually train and adapt the lightweight model at edge with the help of complex model. However, existing analytics approaches leave two challenges untouched: firstly, retraining task is compute-intensive, resulting in large model update delays; secondly, new model may not fit well enough with the data distribution of the current video stream. To address these challenges, in this paper, we present EdgeSync, EdgeSync filters the samples by considering both timeliness and inference results to make training samples more relevant to the current video content as well as reduce the update delay, to improve the quality of training, EdgeSync also designs a training management module that can efficiently adjusts the model training time and training order on the runtime. By evaluating real datasets with complex scenes, our method improves about 3.4% compared to existing methods and about 10% compared to traditional means.

Published
2024

2. Generative Model-based Feature Knowledge Distillation for Action Recognition

Author

Wang, Guiqin, Zhao, Peng, Shi, Yanjiang, Zhao, Cong, and Yang, Shusen

Subjects
Computer Science - Computer Vision and Pattern Recognition
Abstract

Knowledge distillation (KD), a technique widely employed in computer vision, has emerged as a de facto standard for improving the performance of small neural networks. However, prevailing KD-based approaches in video tasks primarily focus on designing loss functions and fusing cross-modal information. This overlooks the spatial-temporal feature semantics, resulting in limited advancements in model compression. Addressing this gap, our paper introduces an innovative knowledge distillation framework, with the generative model for training a lightweight student model. In particular, the framework is organized into two steps: the initial phase is Feature Representation, wherein a generative model-based attention module is trained to represent feature semantics; Subsequently, the Generative-based Feature Distillation phase encompasses both Generative Distillation and Attention Distillation, with the objective of transferring attention-based feature semantics with the generative model. The efficacy of our approach is demonstrated through comprehensive experiments on diverse popular datasets, proving considerable enhancements in video action recognition task. Moreover, the effectiveness of our proposed framework is validated in the context of more intricate video action detection task. Our code is available at https://github.com/aaai-24/Generative-based-KD., Comment: Accepted on AAAI2024

Published
2023

3. Weakly-Supervised Action Localization by Hierarchically-structured Latent Attention Modeling

Author

Wang, Guiqin, Zhao, Peng, Zhao, Cong, Yang, Shusen, Cheng, Jie, Leng, Luziwei, Liao, Jianxing, and Guo, Qinghai

Subjects
Computer Science - Computer Vision and Pattern Recognition
Abstract

Weakly-supervised action localization aims to recognize and localize action instancese in untrimmed videos with only video-level labels. Most existing models rely on multiple instance learning(MIL), where the predictions of unlabeled instances are supervised by classifying labeled bags. The MIL-based methods are relatively well studied with cogent performance achieved on classification but not on localization. Generally, they locate temporal regions by the video-level classification but overlook the temporal variations of feature semantics. To address this problem, we propose a novel attention-based hierarchically-structured latent model to learn the temporal variations of feature semantics. Specifically, our model entails two components, the first is an unsupervised change-points detection module that detects change-points by learning the latent representations of video features in a temporal hierarchy based on their rates of change, and the second is an attention-based classification model that selects the change-points of the foreground as the boundaries. To evaluate the effectiveness of our model, we conduct extensive experiments on two benchmark datasets, THUMOS-14 and ActivityNet-v1.3. The experiments show that our method outperforms current state-of-the-art methods, and even achieves comparable performance with fully-supervised methods., Comment: Accepted to ICCV 2023. arXiv admin note: text overlap with arXiv:2203.15187, arXiv:2003.12424, arXiv:2104.02967 by other authors

Published
2023

33. Immune-related thyroid dysfunction as a positive prognostic factor for patients with lung cancer in China: a real-world retrospective study.

Author

Guo, Ge, Jing, Zihan, Dou, Wenrui, Wang, Guiqin, Dang, JunJie, Li, Yajie, Wang, Ruqiong, Zhang, Huan, Sun, Jing, and Shang, Lihua

Subjects
IMMUNE checkpoint inhibitors, LUNG tumors, PROGNOSIS, CHINESE people, THYROID diseases
Abstract

Introduction: The relationship between immune-related thyroid dysfunction (irTD) and survival rates in cancer patients remains unclear. Furthermore, the impact of variations in immunotherapy line numbers and pathological types among lung cancer patients on this relationship has not been fully elucidated. This study aims to evaluate the potential of irTD as a prognostic marker for immunotherapy in Chinese patients with lung cancer. Methods: A retrospective analysis was conducted on data collected from patients with locally advanced or metastatic lung cancer who received immune checkpoint inhibitor treatment at the Harbin Medical University Cancer Hospital. The study period spanned from December 1, 2016, to November 30, 2023. The primary endpoints were progression-free survival (PFS) and overall survival (OS), while the objective response rate served as the secondary endpoint. Results: Among the 361 patients in this study, 42.7% developed irTD. Significant differences were observed between the groups with and without irTD regarding inflammatory indices, thyroid-stimulating hormone levels, and thyroid autoantibody positivity (P < 0.05). Patients with irTD demonstrated longer OS (32.5 vs. 22 months, HR: 0.65, 95% CI: 0.49-0.88; P = 0.005). For NSCLC patients, OS was significantly prolonged in those with irTD (40.8 vs. 27.2 months, HR: 0.68, 95% CI: 0.48-0.96; P = 0.028). Similarly, SCLC patients who developed irTD exhibited longer OS (27.9 vs. 13.8 months, HR: 0.51, 95% CI: 0.29-0.90; P = 0.022). Notably, irTD was observed exclusively in patients receiving immunotherapy in the second or later lines, showing a significant association with extended OS (40.8 vs. 19.4 months, HR: 0.56, 95% CI: 0.35-0.88; P = 0.012), while the presence of irTD during first-line immunotherapy did not confer a benefit to patients (32.4 vs 24.5 months, HR: 0.74, 95% CI: 0.50-1.10; P = 0.134). The effects of different irTD types, severities, or clinical symptoms on PFS and OS did not differ significantly (P > 0.05). Conclusion: irTD demonstrates potential as a predictive marker for long-term survival benefits in Chinese patients with lung cancer. However, our exploratory analysis indicates that this association was exclusively observed in individuals receiving immunotherapy as a second-line or subsequent treatment. [ABSTRACT FROM AUTHOR]

Published
2025
Full Text
View/download PDF

44. Generation of a Live Attenuated Influenza Vaccine that Elicits Broad Protection in Mice and Ferrets

Author

Wang, Lulan, Liu, Su-Yang, Chen, Hsiang-Wen, Xu, Juan, Chapon, Maxime, Zhang, Tao, Zhou, Fan, Wang, Yao E, Quanquin, Natalie, Wang, Guiqin, Tian, Xiaoli, He, Zhanlong, Liu, Longding, Yu, Wenhai, Sanchez, David Jesse, Liang, Yuying, Jiang, Taijiao, Modlin, Robert, Bloom, Barry R, Li, Qihan, Deng, Jane C, Zhou, Paul, Qin, F Xiao-Feng, and Cheng, Genhong

Subjects
Medical Microbiology, Biomedical and Clinical Sciences, Immunology, Vaccine Related, Infectious Diseases, Pneumonia & Influenza, Influenza, Immunization, Emerging Infectious Diseases, Prevention, Biodefense, Biotechnology, 3.4 Vaccines, 2.1 Biological and endogenous factors, Infection, Adoptive Transfer, Animals, Cross Protection, DNA Transposable Elements, Disease Models, Animal, Ferrets, Genetic Testing, Immunity, Heterologous, Influenza A Virus, H1N1 Subtype, Influenza A Virus, H3N2 Subtype, Influenza A Virus, H5N1 Subtype, Influenza Vaccines, Mice, Mutagenesis, Insertional, Orthomyxoviridae Infections, Survival Analysis, T-Lymphocytes, Vaccines, Attenuated, flu vaccine, genome-wide mutagenesis, influenza vaccine, matrix protein, Microbiology, Biochemistry and cell biology, Medical microbiology
Abstract

New influenza vaccines that provide effective and broad protection are desperately needed. Live attenuated viruses are attractive vaccine candidates because they can elicit both humoral and cellular immune responses. However, recent formulations of live attenuated influenza vaccines (LAIVs) have not been protective. We combined high-coverage transposon mutagenesis of influenza virus with a rapid high-throughput screening for attenuation to generate W7-791, a live attenuated mutant virus strain. W7-791 produced only a transient asymptomatic infection in adult and neonatal mice even at doses 100-fold higher than the LD50 of the parent strain. A single administration of W7-791 conferred full protection to mice against lethal challenge with H1N1, H3N2, and H5N1 strains, and improved viral clearance in ferrets. Adoptive transfer of T cells from W7-791-immunized mice conferred heterologous protection, indicating a role for T cell-mediated immunity. These studies present an LAIV development strategy to rapidly generate and screen entire libraries of viral clones.

Published
2017

49. Inhibition of Staphylococcus aureus biofilm by quercetin combined with antibiotics.

Author

Ma, Yanjun, Kang, Xinyun, Wang, Guiqin, Luo, Shuangyan, Luo, Xiaofeng, and Wang, Guilai

Subjects
LASER microscopy, CONFOCAL microscopy, GENTIAN violet, SCANNING electron microscopy, STAPHYLOCOCCUS aureus, QUERCETIN, TETRACYCLINES
Abstract

This study aimed to investigate the effects of combined quercetin and antibiotics on the bacteriostatic activity and biofilm formation of Staphylococcus aureus. Optimal concentrations of quercetin and antibiotics (tetracycline and doxycycline) for inhibiting biofilm formation were determined using the Fractional Inhibitory Concentration Index and Minimum Biofilm Inhibitory Concentration assays. The impact of the drug combinations on biofilm clearance at various formation stages was determined using crystal violet staining, scanning electron microscopy and confocal laser microscopy. The results indicated that quercetin enhanced the bactericidal effect of tetracycline antibiotics against S. aureus. The combination significantly reduced both the metabolic activity within S. aureus biofilms and the production of biofilm matrix components. Scanning electron microscopy and confocal laser microscopy confirmed that the combination treatment significantly reduced bacterial cell counts within the biofilm. Quercetin treatment significantly increased the sensitivity of biofilms to antibiotics, supporting its potential application as a novel antibiotic synergist. [ABSTRACT FROM AUTHOR]

Published
2024
Full Text
View/download PDF

Catalog

Books, media, physical & digital resources
See catalog results