A5082334473- Visual perception and processing mechanisms
- Color perception and design
- Ocular and Laser Science Research
- Color Science and Applications
- Computational Drug Discovery Methods
- Glaucoma and retinal disorders
- Impact of Light on Environment and Health
- Engineering Applied Research
- Machine Learning in Materials Science
- Infrastructure Maintenance and Monitoring
- Photonic and Optical Devices
- Urinary Bladder and Prostate Research
- Statistical and Computational Modeling
- Advanced Optical Imaging Technologies
- Gaze Tracking and Assistive Technology
- X-ray Diffraction in Crystallography
- Wind and Air Flow Studies
- Prostate Cancer Diagnosis and Treatment
- Advanced Neural Network Applications
- Fuel Cells and Related Materials
- Energy and Environmental Systems
- Irrigation Practices and Water Management
- Bioinformatics and Genomic Networks
- Domain Adaptation and Few-Shot Learning
- Advanced Fluorescence Microscopy Techniques
Beijing University of Technology
2023-2025
Beijing Tsinghua Chang Gung Hospital
2025
Tsinghua University
2025
Fudan University
2025
Southeast University
2022-2024
Tongji University
2024
Yangzhou University
2023
The oxygen reduction reaction (ORR) and the evolution (OER) are crucial for conversion of clean energy. Recently, dual-metal-site catalysts (DMSCs) have gained much attention due to their high atom utilization, stronger stability, better catalytic performance. An advanced method that combines density functional theory (DFT) machine learning (ML) has been employed in this study investigate adsorption free energies adsorbates on hundreds potential catalysts, with aim screening highly active...
In the realm of drug discovery and design, accurate prediction protein–ligand binding affinity is paramount importance as it underpins functional interactions within biological systems. This study introduces a novel self-supervised learning (SSL) framework that combines contrastive graph neural networks (CL-GNN) for predicting affinities, which critical aspect discovery. Traditional methods are expensive time-consuming, prompting development more efficient computational approaches. CL-GNN...
Abstract BACKGROUND Rice cultivation under film mulching with no flooding is widely used as an effective water‐saving technology. Different colors of mulch have different effects on the soil hydrothermal environment and crop growth because their optical properties. However, temperature rice physiological are not clearly understood. RESULTS Field experiments were conducted in 2019 2020 to investigate color mulches a non‐flooded condition. Transparent (TM), black (BM), two‐color (BWM, silver...
Recently, in the field of crystal property prediction, graph neural network (GNN) model has made rapid progress. The GNN can effectively capture high-dimensional features from structure, thereby achieving optimal performance prediction. However, existing faces limitations handling hidden layer after pooling layer, which restricts training model. In present research, we propose a novel called batch normalization multilayer perceptron distance (BNM-CDGNN). BNM-CDGNN encodes crystal's geometry...
Abstract As an important index in studies on vision and display technology, pupil size controls the amount of light entering human eye. Eye tracking systems are normally used to measure while a estimation model can be applied when actual measurement is not available. The weighting functions reported models show radial symmetry, which contradicted asymmetric distribution photoreceptors retina. Chromaticity spatial has been taken into consideration either. In this paper, perceptual experiments...
In augmented reality diffractive waveguide technology, the light field needs to be collimated before being transmitted into waveguide. Conventional schemes usually require additional collimating optics collimate from micro-image source and guide it in-coupling elements. order meet of head-mounted devices further miniaturize equipment, this paper proposes a device that combines collimation coupling by using reflective polarization volume lens (PVL). A related model is also established...
The success of pretrain-finetune paradigm brings about the release numerous model weights. In this case, merging models finetuned on different tasks to enable a single with multi-task capabilities is gaining increasing attention for its practicability. Existing methods usually suffer from (1) significant performance degradation or (2) requiring tuning by additional data training. paper, we rethink and analyze existing paradigm. We discover that using model's weights can hardly simulate all...
Through psychophysical experiments, the study investigated effects of ambient illuminance and screen luminance on perceived brightness pupil diameter. The results indicate a close correlation between both illuminance. changes in diameter reflect to some extent perception. A preliminary perception model was established quantitatively predict how human eye perceives luminance.
The mismatch between physical brightness and perceived of display devices in different light environments is increasingly serious. This paper analyzes the perception mechanism according to structure characteristics human visual system. Different stimuli were further designed for experiments, influence relevant objective quantities on evaluation was studied, so as establish a model based retinal imaging. Ultimately, can be characterized by luminance pupil diameter, it has high correlation...
This study retrospectively examines the effectiveness of low-frequency electrical stimulation for addressing nocturia in patients experiencing urinary incontinence after undergoing radical prostatectomy. We reviewed outcomes 32 who had undergone prostatectomy and subsequently experienced incontinence. These were divided into control group ( n = 16) received pelvic floor muscle training treatment underwent electrophysiological appropriate technique conjunction with training. assessed changes...
People's perception of brightness and visual comfort in different light environments are important indicators display products. The the human eye is often from actual display. This because there a difference between perceived physical brightness. luminous body same may give people feelings, thus producing In complex optical environment, existing photoelectric measurement parameters not be able to accurately describe effect device. research on has long history, but view increasingly...
We report a low‐cost, high‐performance and deep‐learning-based passive Non‐Line‐of‐Sight (NLOS) imaging method. This method applies the pix2pix network to reconstruct NLOS scenes reconstructed images show high quality. Our could promote realization of displaying on head-mounted AR displays because quality reconstruction speed, thus be applied fields including anti-terrorism autonomous driving where both speed image
This work targets to merge various Vision Transformers (ViTs) trained on different tasks (i.e., datasets with object categories) or domains the same categories but environments) into one unified model, yielding still good performance each task domain. Previous model merging works focus either CNNs NLP models, leaving ViTs research untouched. To fill this gap, we first explore and find that existing methods cannot well handle of whole ViT models have improvement space. enable ViT, propose a...
As an important parameter in studies on vision and ophthalmology, pupil size directly controls the amount of light entering human eye. Pupilometers or eye tracking systems are used to measure some while a estimation model can be applied when actual measurement is not available, such as cases optical design, photobiological safety studies, etc. There several existing models, most which reveal relationship between spatial distribution luminance. However, radial asymmetry luminance weighting...