A5100676828- Anomaly Detection Techniques and Applications
- Adversarial Robustness in Machine Learning
- Air Quality Monitoring and Forecasting
- Privacy-Preserving Technologies in Data
- Topic Modeling
- Seismology and Earthquake Studies
- Cryptography and Data Security
- Machine Learning and Data Classification
- Advanced Vision and Imaging
- Network Security and Intrusion Detection
- Internet Traffic Analysis and Secure E-voting
- Advanced SAR Imaging Techniques
- Energy Load and Power Forecasting
- Pulsars and Gravitational Waves Research
- Machine Learning and ELM
- Stochastic Gradient Optimization Techniques
- Biometric Identification and Security
- Remote Sensing and LiDAR Applications
- Text and Document Classification Technologies
- Robot Manipulation and Learning
- Cell Image Analysis Techniques
- Kruppel-like factors research
- Advanced Image and Video Retrieval Techniques
- Atmospheric chemistry and aerosols
- Stock Market Forecasting Methods
Fudan University
2011-2024
Northeast Petroleum University
2024
China University of Mining and Technology
2024
Center for Drug Evaluation and Research
2020-2022
University of Chinese Academy of Sciences
2021
Qihoo 360 (China)
2021
United States Food and Drug Administration
2020
Ocean University of China
2019
Université de Montpellier
2019
Centre National de la Recherche Scientifique
2019
Activation of β-catenin, the central effector canonical wingless-type (Wnt) pathway, has been implicated in hepatocellular carcinoma (HCC). However, transcription regulation mechanism β-catenin gene HCC remains unknown. Here we report that human zinc finger protein 191 (ZNF191) is a potential regulator transcription. ZNF191, Krüppel-like protein, specifically interacts with TCAT motif, which constitutes HUMTH01 microsatellite tyrosine hydroxylase ( TH ) ex vivo . We demonstrate ZNF191...
Effective chart summary can significantly reduce the time and effort decision makers spend interpreting charts, enabling precise efficient communication of data insights. Previous studies have faced challenges in generating accurate semantically rich summaries time-series charts. In this paper, we identify elements common hallucination types generation summaries, which serve as our guidelines for automatic generation. We introduce ChartInsighter, automatically generates data, effectively...
Most photometric stereo approaches assume distant or directional lighting and orthographic imaging. However, when the source is divergent near object camera projective, image intensity of a Lambertian non-linear function both unknown surface normals distances to points. The resulting optimization non-convex highly sensitive initial guess. In this paper, we propose two-stage near-light method using circularly placed point light sources (commonly seen in recent consumer imaging devices like...
Multiple object tracking (MOT) has been successfully investigated in computer vision. However, MOT for the videos captured by unmanned aerial vehicles (UAV) is still challenging due to small size, blurred appearance, and very large and/or irregular motion both ground objects UAV platforms. In this paper, we propose FOLT mitigate these problems reach fast accurate view. Aiming at speed-accuracy trade-off, adopts a modern detector light-weight optical flow extractor extract detection features...
In this paper we present DELTA, a deep learning based language technology platform. DELTA is an end-to-end platform designed to solve industry level natural and speech processing problems. It integrates most popular neural network models for training as well comprehensive deployment tools production. aims provide easy fast experiences using, deploying, developing both academia use cases. We demonstrate the reliable performance with on several tasks, including text classification, named...
ASR (Automatic Speech Recognition) technology is a key for human-computer interaction. Especially the DNN models of wake-up-word speech recognition, which enables smart device to recognize wake-up words spoken by users when they are in sleep or lock screen state, allowing directly enter wait command and start first step voice no wonder provides great convenience people's daily life, but it's security problem has always been hot topic further research. The rapid development made these very...
Detecting the abnormal behavior of network through measurement is very important for security. In high-speed networks, it difficult and unrealistic to perform per-packet analysis with limited computing resources. Packet sampling can greatly reduce consumption this paper, we compare five methods based on from theory experiment, which are mean method independent (mean-IS), median (median-IS), dependent (mean-DS), (median-DS), random (RP), respectively. We simulate process by Monte Carlo...
The recent successful detection of gravitational waves (GWs) at nanohertz based on pulsar timing arrays has underscored the growing significance searching for new pulsars, which serve as valuable probes GWs. However, one challenges in this endeavor is lack labeled data, can lead to overfitting and poor generalization supervised deep neural networks. In paper, we propose a self-supervised pretext task signal con-texts obtain discriminative radio representation. Specially, attentions are...
Abstract Accurate spatiotemporal estimation of non‐methane volatile organic compounds (NMVOCs) plays a pivotal role in establishing sophisticated early warning systems and formulating strategies to combat air pollution. Despite these critical applications, robust high resolution NMVOCs concentrations remains challenge. In this study, we develop space‐time Light Gradient Boosting Machine (STLGB) model, which successfully renders hourly maps across Shanghai from 2019 2022 by integrating...
In this paper, we propose a scalable clustering paradigm to address the problems of excessive computational load and limited performance in large-scale data. The proposed method employs enhanced splitting merging awareness tactics (E-SMART) algorithm. dataset is divided into many sub-datasets sampled randomly from original These are clustered using E-SMART with number clusters K detected automatically resulting partitions combined re-clustered. We evaluate our synthetic fMRI datasets...
In 1998, Tseng and Jan proposed a lightweight interactive user identification protocol based on ID-based cryptography. Recently, Hwang et al. modified their to reduce the responding waiting time for wireless network applications. this letter, we show that scheme is vulnerable impersonation attacks.
The problem of signal detection and classification multiple UAVs can be solved using object techniques in computer vision. However, this requires collecting labeling a large amount reliable raw data. Since the UAV dataset cannot directly applied to detection, we propose method time-frequency domain filtering automatic construct large-scale spectrogram dataset. Experimental results show that average recognition accuracies image transmission signals remote control under interference conditions...
Stochastic Gradient Descent (SGD), a widely used optimization algorithm in deep learning, is often limited to converging local optima due the non-convex nature of problem. Leveraging these improve model performance remains challenging task. Given inherent complexity neural networks, simple arithmetic averaging obtained models undesirable results. This paper proposes {\em soft merging} method that facilitates rapid merging multiple models, simplifies specific parts and enhances robustness...
In this paper, a method for selecting neural network samples is proposed, which to construct sample training multivariate model by set of data points with high correlation degree. Choosing the right has great influence on prediction accuracy final model, and it also an important task forecasters. temperature anomaly ocean subsurface predicted Convolution Neural Network (CNN) CNN uses Surface Correlation Degree Method (SCDM) proposed in paper blank area zero padding select training, so as...
Adversarial attacks attempt to disrupt the training, retraining and utilizing of artificial intelligence machine learning models in large-scale distributed systems. This causes security risks on its prediction outcome. For example, attackers poison model by either presenting inaccurate misrepresentative data or altering models' parameters. In addition, Byzantine faults including software, hardware, network issues occur systems which also lead a negative impact this paper, we propose novel...