A5100402302- Privacy-Preserving Technologies in Data
- Adversarial Robustness in Machine Learning
- Cryptography and Data Security
- Stochastic Gradient Optimization Techniques
- Neural Networks and Applications
- Biometric Identification and Security
- Machine Learning in Healthcare
- Advanced Memory and Neural Computing
- Rough Sets and Fuzzy Logic
- Face recognition and analysis
- Data Quality and Management
- Advanced Steganography and Watermarking Techniques
- Distributed Sensor Networks and Detection Algorithms
- Reinforcement Learning in Robotics
- Data Management and Algorithms
- Sarcoidosis and Beryllium Toxicity Research
- Anomaly Detection Techniques and Applications
- Internet Traffic Analysis and Secure E-voting
- Advanced Computational Techniques and Applications
- Face and Expression Recognition
- Semantic Web and Ontologies
- Vasculitis and related conditions
- Network Security and Intrusion Detection
- Advanced Neural Network Applications
- Advanced Malware Detection Techniques
Beijing University of Technology
2025
Harbin Normal University
2011-2024
Sichuan Agricultural University
2024
Beijing University of Posts and Telecommunications
2019-2024
Hangzhou Dianzi University
2024
West China Medical Center of Sichuan University
2024
California Institute for Biomedical Research
2023
First Bethune Hospital of Jilin University
2023
First Affiliated Hospital of Zhengzhou University
2002-2022
Huazhong Agricultural University
2022
Recent years have witnessed a growing interest in developing automatic palmprint recognition methods. Among them, coding-based ones, representing the texture of using binary code, are most prevalent and successful. We find that not all bits code map generated by specific coding scheme equally consistent. A bit is deemed fragile if its value changes across maps created from different images same palmprint. In this paper, we first analyze phenomenon state-of-the-art scheme, namely, orientation...
Developing 3D palmprint recognition systems has recently begun to draw attention of researchers. Compared with its 2D counterpart, several unique merits. However, most the existing matching methods are designed for one-to-one verification and they not efficient cope one-to-many identification case. In this paper, we fill gap by proposing a collaborative representation (CR) based framework l1-norm or l2-norm regularizations identification. The effects different regularization terms have been...
This study undertakes a comprehensive examination of the intricate link between diet nutrition, age, and metabolic syndrome (MetS), utilizing advanced artificial intelligence methodologies. Data from National Health Nutrition Examination Survey (NHANES) spanning 1999 to 2018 were meticulously analyzed using machine learning (ML) techniques, specifically extreme gradient boosting (XGBoost) proportional hazards model (COX). Using these analytic methods, we elucidated significant correlation...
As a kind of prevalent malignancy globally, hepatocellular carcinoma (HCC) is characterized by significant morbidity and mortality due to the difficulties in early diagnosis limited treatment options. Circular RNAs (circRNAs) are type circular single-stranded RNA molecule formed back-splicing 5' end 3' linear RNA, possessing multiple biological functions. In recent years, numerous reports have demonstrated that circRNAs potential biomarkers therapeutic targets for HCC. this study, we found...
Federated learning is a distributed method to train shared model by aggregating the locally-computed gradient updates. In federated learning, bandwidth and privacy are two main concerns of updates transmission. This paper proposes an end-to-end encrypted neural network for first encodes input lower-dimension space in each client, which significantly mitigates pressure data communication learning. The encoded directly recovered as whole, i.e. aggregated trained model, decoding layers on...
There has been an increase of interest in code search using natural language. Assessing the performance such models can be difficult without a readily available evaluation suite. In this paper, we present dataset consisting language query and snippet pairs, with hope that future work area use as common benchmark. We also provide results two ([1] [6]) from recent work. The is at https://github.com/facebookresearch/Neural-Code-Search-Evaluation-Dataset
Abstract Background Numerous of models have been developed to predict the bone metastasis (BM) risk; however, due variety cancer types, it is difficult for clinicians use these efficiently. We aimed perform pan‐cancer analysis create classification system BM, and construct nomogram predicting BM risk. Methods Cancer patients diagnosed between 2010 2018 in Surveillance, Epidemiology, End Results (SEER) database were included. Unsupervised hierarchical clustering was performed prevalence‐based...
Our objective was to determine metabolic syndrome (MS) prevalence in Chinese patients with systemic lupus erythematosus (SLE) and investigate the conditions that contribute its development. 116 SLE classified according American College of Rheumatology (ACR) classification criteria, 115 controls were enrolled. MS defined by joint interim statement International Diabetes Federation Task Force on Epidemiology Prevention; National Heart, Lung, Blood Institute; Heart Association; World...
AI is making the Web an even cooler place, but also introduces serious privacy risks due to extensive user data collection. Federated learning (FL), as a privacy-preserving machine paradigm, enables mobile devices collaboratively learn shared prediction model while keeping all training on devices. However, key obstacle towards practical cross-device FL huge energy consumption, especially for lightweight In this work, we perform first-of-its-kind analysis of improving performance through...
Federated learning enables collaborative machine among multiple independent participants while preserving data privacy of each participant through model aggregation during training. However, still faces potential risks that are associated with indirect leakage, such as parameters. In this paper, we first propose an algorithm called Meta Operation Neural Network (MONN) to perform basic arithmetic operations on encrypted and generate operation results in a plaintext way. MONN is actually...
With strict protections and regulations of data privacy security, conventional machine learning based on centralized datasets is confronted with significant challenges, making artificial intelligence (AI) impractical in many mission-critical data-sensitive scenarios, such as finance, government, health. In the meantime, tremendous are scattered isolated silos various industries, organizations, different units an organization, or branches international organization. These valuable resources...
Medical assisted decision-making plays a key role in providing accurate and reliable medical advice. But decision-making, various uncertainties are often accompanied. The belief rule base (BRB) has strong nonlinear modeling capability can handle well. However, BRB suffers from combinatorial explosion tends to influence explainability during the optimization process. Therefore, an interval with (IBRB-e) is explored this paper. Firstly, pre-processing using extreme gradient boosting (XGBoost)...
ABSTRACT The state‐of‐health (SOH) assessment of lithium‐ion batteries is critical to the development and optimization maintenance strategies. To ensure accuracy results, it must not only address a variety uncertainties but also rationalize transparently conduct process, as well make results interpretable traceable. These requirements are necessary that battery operates safely steadily. As an modeling method, belief rule base (BRB) has been widely used in SOH assessment. However, current...
Federated learning is a distributed method to train shared model by aggregating the locally-computed gradient updates. In federated learning, bandwidth and privacy are two main concerns of updates transmission. This paper proposes an end-to-end encrypted neural network for first encodes input lower-dimension space in each client, which significantly mitigates pressure data communication learning. The encoded directly recovered as whole, i.e. aggregated trained model, decoding layers on...
Accurate network traffic classification is of urgent need in the big data era, as anomalous becomes formidable to classify nowadays complicated environment. Deep Learning (DL) techniques can master detecting due capability fitting training data. However, this lay on correctness data, which also made them sensitive annotation errors. We propose that by measuring uncertainty model, errors be accurately corrected for classifying traffic. use dropout approximate prior distribution and calculate...