A5033410733- Biometric Identification and Security
- Adversarial Robustness in Machine Learning
- Face and Expression Recognition
- Face recognition and analysis
- Anomaly Detection Techniques and Applications
- Wastewater Treatment and Nitrogen Removal
- Network Security and Intrusion Detection
- Forensic and Genetic Research
- Advanced Neural Network Applications
- Industrial Vision Systems and Defect Detection
- Digital Media Forensic Detection
- Artificial Intelligence in Healthcare
- Image and Video Stabilization
- Water Quality Monitoring and Analysis
- Artificial Immune Systems Applications
- Advanced Chemical Sensor Technologies
- Advanced Steganography and Watermarking Techniques
- Retinal Diseases and Treatments
- Retinal Imaging and Analysis
- Advanced Computational Techniques and Applications
- Complex Network Analysis Techniques
- Forensic Fingerprint Detection Methods
- Spectroscopy and Chemometric Analyses
- Generative Adversarial Networks and Image Synthesis
- Machine Learning in Healthcare
Hefei University of Technology
2025
Changchun University of Technology
2024
Tsinghua University
2011-2024
Lanzhou University
2024
Qingdao Binhai University
2024
Shandong University
2024
State Key Laboratory of Pollution Control and Resource Reuse
2024
Nanjing University
2024
Gansu Meteorological Bureau
2023
Shandong Jianzhu University
2022-2023
Face anti-spoofing is an important task in full-stack face applications including detection, verification, and recognition. Previous approaches build models on datasets which do not simulate the real-world data well (e.g., small scale, insignificant variance, etc.). Existing may rely auxiliary information, prevents these solutions from generalizing practice. In this paper, we present a collection solution along with synthesis technique to digital medium-based spoofing attacks, can easily...
Deep neural networks are vulnerable to adversarial examples, which becomes one of the most important research problems in development deep learning. While a lot efforts have been made recent years, it is great significance perform correct and complete evaluations attack defense algorithms. In this paper, we establish comprehensive, rigorous, coherent benchmark evaluate robustness on image classification tasks. After briefly reviewing plenty representative methods, large-scale experiments...
Face recognition has been an active and vital topic among computer vision community for a long time. Previous researches mainly focus on loss functions used facial feature extraction network, which the improvements of softmax-based greatly promote performance face recognition. However, contradiction between drastically increasing number identities shortage GPU memory is gradually becoming irreconcilable. In this work, we theoretically analyze upper limit model parallelism in first place....
Network intrusion detection is an important technology in national cyberspace security strategy and has become a research hotspot various issues recent years. The development of effective efficient intelligent network methods using advanced machine learning algorithms great importance for defending against intrusions complex environments. In this study, method based on decision tree twin support vector hierarchical clustering, named HC-DTTWSVM, proposed, which can effectively detect...
Dictionary based orientation field estimation approach has shown promising performance for latent fingerprints. In this paper, we seek to exploit stronger prior knowledge of fingerprints in order further improve the performance. Realizing that ridge orientations at different locations have characteristics, propose a localized dictionaries-based algorithm, which noisy patch location output by local is replaced real dictionary same location. The precondition applying dictionaries pose...
Anomaly detection aims to identify deviations from normal patterns within data. This task is particularly crucial in dynamic graphs, which are common applications like social networks and cybersecurity, due their evolving structures complex relationships. Although recent deep learning-based methods have shown promising results anomaly on they often lack of generalizability. In this study, we propose GeneralDyG, a method that samples temporal ego-graphs sequentially extracts structural...
Activated sludge (AS) bulking is a significant challenge in AS processes, and therefore, predicting the settling performance of essential to maintaining long-term stable operation wastewater treatment plants (WWTPs). In this study, samples taken from 42 WWTPs three laboratory reactors were imaged labeled with volume indexes predict sedimentation based on deep learning models. A tagged image database was established 105,695 images. Comparing five different algorithms suggested that...
Face recognition has been an active and vital topic among computer vision community for a long time. Previous researches mainly focus on loss functions used facial feature extraction network, which the improvements of softmax-based greatly promote performance face recognition. However, contradiction between drastically increasing number identities shortage GPU memories is gradually becoming irreconcilable. In this paper, we thoroughly analyze optimization goal difficulty training massive...
Fluorescence spectroscopy attracted more and attention in pesticide residue detection field because of its advantages non-destructive, non-contact, high speed no requirement complex pre-process procedure. However, given that the concentration detected via fluorescence is calculated accordance with Beer-Lambert law, this method can only be used to detect samples containing a single kind or several kinds pesticides completely different which not practical cases. In article, overcome...
Healthcare spending has been increasing in the last few decades. One of main reasons for this increase is hospital readmissions, which defined as a re-hospitalization patient after being discharged from within short period time. The excessive amount money spent every year on readmissions and urge to enhance healthcare quality make reducing necessity. In paper, we extract knowledge medical dataset apply concept mining actionable rules guide health domain experts their decision-making process....
In recent years, law enforcement agencies are increasingly using palmprint to identify criminals. For identification systems, efficiency is a very important but challenging problem because of large database size and poor image quality. Existing systems not sufficiently fast for practical applications. To solve this problem, novel indexing algorithm based on ridge features proposed in paper. A pre-aligned by registering its orientation field with respect set reference fields, which obtained...
AIM: To explore the correlation between cystatin C (Cys-C) and diabetic retinopathy (DR) in those patients with type 2 diabetes mellitus (DM) China. METHODS: Articles were collected from China National Knowledge Infrastructure (CNKI), Wanfang, VIP, PubMed, EMBASE, Cochrane Library, Clinical Trials.gov, Google Scholar. Quality risk of bias within included studies was assessed using Newcastle-Ottawa scale (NOS). Heterogeneity determined by Cochran’s Q-test Higgins I2 statistics. Mean...
Latent fingerprints have been used by law enforcement agencies to identify suspects for a century. However, because of poor image quality and complex background noise, latent are routinely identified relying on features manually marked human experts in practice. A large number can not be treated time due lacking well trained experts, highlighting the need "lights out" (fully-automatic) systems. In this paper, we propose systematic algorithm fingerprint detection, segmentation, orientation...
To explore the association between metabolic syndrome (MetS) and its component thyroid volume in Chinese adolescents, to compare detection rate of MetS under three different diagnostic criteria.A total 1097 school students (610 males 487 females, ages 12-15 years) were enrolled. All participants underwent physical examination, biochemical test, gland ultrasonography. The normal, overweight obese group was compared. We also analyzed number components volume. Linear multiple linear regression...
Heterogeneous face recognition (HFR), referring to matching images across different domains, is a challenging problem due the vast cross-domain discrepancy and insufficient pairwise training data. This article proposes quadruplet framework for learning domain-invariant discriminative features (DIDF) HFR, which integrates domain-level class-level alignment in one unified network. The reduces distribution discrepancy. based on special loss developed further diminish intra-class variations...
Many works have investigated the adversarial attacks or defenses under settings where a bounded and imperceptible perturbation can be added to input. However in real-world, attacker does not need comply with this restriction. In fact, more threats deep model come from unrestricted examples, that is, makes large visible modifications on image, which causes classifying mistakenly, but affect normal observation human perspective. Unrestricted attack is popular practical direction has been...