A5100724377- Video Surveillance and Tracking Methods
- Particle Accelerators and Free-Electron Lasers
- Particle accelerators and beam dynamics
- Radiomics and Machine Learning in Medical Imaging
- Retinal Imaging and Analysis
- Image Enhancement Techniques
- EEG and Brain-Computer Interfaces
- Advanced Image Processing Techniques
- AI in cancer detection
- Medical Image Segmentation Techniques
- Human Pose and Action Recognition
- Image and Signal Denoising Methods
- Colorectal Cancer Screening and Detection
- Retinal and Optic Conditions
- Gaze Tracking and Assistive Technology
- Innovative Teaching and Learning Methods
- Online Learning and Analytics
- Face recognition and analysis
- Advanced Vision and Imaging
- Photoacoustic and Ultrasonic Imaging
- Adversarial Robustness in Machine Learning
- Glaucoma and retinal disorders
- Intelligent Tutoring Systems and Adaptive Learning
- Impact of Light on Environment and Health
- Gyrotron and Vacuum Electronics Research
Shandong Normal University
2019-2024
Guilin University of Electronic Technology
2024
Northeastern University
2013-2020
First Affiliated Hospital of Jinan University
2020
Goddard Space Flight Center
2018
Guangdong General Hospital
2018
Guangdong Academy of Medical Sciences
2018
South China University of Technology
2018
Shandong University
2015-2017
National Sun Yat-sen University
2017
In recent years, deep learning has shown very competitive performance in seizure detection. However, most of the currently used methods either convert electroencephalogram (EEG) signals into spectral images and employ 2D-CNNs, or split one-dimensional (1D) features EEG many segments 1D-CNNs. Moreover, these investigations are further constrained by absence consideration for temporal links between time series spectrogram images. Therefore, we propose a Dual-Modal Information Bottleneck...
A wide range of defenses have been proposed to harden neural networks against adversarial attacks. However, a pattern has emerged in which the majority are quickly broken by new Given lack success at generating robust defenses, we led ask fundamental question: Are attacks inevitable? This paper analyzes examples from theoretical perspective, and identifies bounds on susceptibility classifier We show that, for certain classes problems, inescapable. Using experiments, explore implications...
Medical image reconstruction methods based on deep learning have recently demonstrated powerful performance in photoacoustic tomography (PAT) from limited-view and sparse data. However, because most of these must utilize conventional linear to implement signal-to-image transformations, their is restricted. In this paper, we propose a novel approach that integrates appropriate data pre-processing training strategies. The Feature Projection Network (FPnet) presented herein designed learn...
Accurate segmentation of brain tumors in MRI images is imperative for precise clinical diagnosis and treatment. However, existing medical image methods exhibit errors, which can be categorized into two types: random errors systematic errors. Random arising from various unpredictable effects, pose challenges terms detection correction. Conversely, attributable to effectively addressed through machine learning techniques. In this paper, we propose a corrective diffusion model accurate tumor by...
While many seizure detection methods have demonstrated great accuracy, their training necessitates a substantial volume of labeled data. To address this issue, we propose novel method for unsupervised anomaly called SAnoDDPM, which uses denoising diffusion probabilistic models (DDPM). We designed pipeline that variable lower bound on Markov chains to identify potential values are unlikely occur in anomalous The model is first trained normal data, then data input the model. resamples and...
An efficient machine learning scheme using a SVM classifier for predicting the aggregation-induced emission effect of triphenylamine-based luminophores was proposed.
We observe a common characteristic between the classical propagation-based image matting and Gaussian process (GP)-based regression. The former produces closer alpha matte values for pixels associated with higher affinity, while outputs regressed by latter are more correlated similar inputs. Based on this observation, we reformulate as GP find that novel matting-GP formulation results in set of attractive properties. First, it offers an alternative view approach to matting. Second,...
Abstract Background Recently, deep convolutional neural networks (CNNs) have been widely adopted for ultrasound sequence tracking and shown to perform satisfactorily. However, existing trackers ignore the rich temporal contexts that exists between consecutive frames, making it difficult these perceive information about motion of target. Purpose In this paper, we propose a sophisticated method fully utilize sequences with bottleneck. This determines frames both feature extraction similarity...
Contrast-enhanced spectral mammography (CESM) is an effective tool for diagnosing breast cancer with the benefit of its multiple types images. However, few models simultaneously utilize this feature in deep learning-based classification methods. To combine features CESM and thus aid physicians making accurate diagnoses, we propose a hybrid approach by taking advantages both fusion models.We evaluated proposed method on dataset obtained from 95 patients between ages ranging 21 to 74 years,...
In visual tracking, a mature scale estimation method can greatly improve tracking performance and provide accurate target information for model training. However, many approaches ignore the problem or adopt heuristic exhaustive scale-estimation strategy. this paper, we propose novel correlation-filter based approach that reveals missing link between detection response. contrast to multi-scale trackers, which generate samples at different scales using some pre-designed criteria then select...
Object tracking is an important capability for robots tasked with interacting humans and the environment, it enables to manipulate objects. In object tracking, selecting samples learn a robust efficient appearance model challenging task. Model learning determines both strategy frequency of updating, which concerns many details that can affect results. this paper, we propose approach by formulating new objective function integrates paradigm self-paced into such reliable be automatically...
In visual tracking, learning a robust and efficient appearance model is challenging task. Model determines both the strategy frequency of updating, which contains many details that could affect tracking results. Self-paced (SPL) has recently been attracting considerable interest in fields machine computer vision. SPL inspired by principle underlying cognitive process humans, whose generally from easier samples to more complex aspects We propose method integrates paradigm into so reliable can...
Visual tracking is a fundamental capability for robots tasked with humans and environment interaction. However, state-of-the-art visual methods are still prone to failures imprecise when applied challenging stereos, their results generally confidence agonistic. These depend on an embedded deep learning model provide deterministic features or regression maps. A output low can result in disastrous consequences lacks evidence needed subsequent operations. Moreover, training data ambiguities...
Xi an Proton Application Facility (XiPAF) which consists of one 230 MeV proton accelerator and irradiation stations, will be constructed in city, Shaanxi, China. The facility is composed a synchrotron, 7 H⁻ linac injector two experimental stations. It can provide flux 10⁵~10⁸ p/cm²/s with the uniformity better than 90% on 10 cm×10 cm sample. overall design XiPAF presented this paper. And progress project reported also.