A5025002931- Pulsars and Gravitational Waves Research
- Advanced Image and Video Retrieval Techniques
- Power Systems and Technologies
- Cell Image Analysis Techniques
- Human Pose and Action Recognition
- Color Science and Applications
- Hand Gesture Recognition Systems
- Rice Cultivation and Yield Improvement
- Smart Grid Energy Management
- Online Learning and Analytics
- Image Processing Techniques and Applications
- Integrated Energy Systems Optimization
- Multilevel Inverters and Converters
- Advanced Battery Technologies Research
- Speech and Audio Processing
- Advanced DC-DC Converters
- Silicon Effects in Agriculture
- Computer Graphics and Visualization Techniques
- AI in cancer detection
- Visual Attention and Saliency Detection
- Surface Modification and Superhydrophobicity
- Geophysics and Gravity Measurements
- Astrophysical Phenomena and Observations
- Music and Audio Processing
- Speech and dialogue systems
Tencent (China)
2023-2024
Fudan University
2023-2024
China Agricultural University
2021-2023
Guiyang University
2022
Energy Research Institute
2021-2022
Queen's University
2020
The explosive growth of online review data on e-commerce websites demands major changes to product emotional design. Kansei engineering, the primary approach design, faces multiple problems. This study aims improve engineering method adapt design trends big era. Targeting reviews, authors mined text data, calculated fine-grained emotions, and presented an innovative flow for engineering. Firstly, a crawler program was compiled mine reviews parameters from websites, were processed. Next, word...
Convolution-based and Transformer-based vision backbone networks process images into the grid or sequence structures, respectively, which are inflexible for capturing irregular objects. Though Vision GNN (ViG) adopts graph-level features complex images, it has some issues, such as inaccurate neighbor node selection, expensive information aggregation calculation, over-smoothing in deep layers. To address above problems, we propose a Progressive Graph (PVG) architecture recognition task....
Separation of oil from emulsions is industrially important. Existing Janus filters that bear a de-emulsifier on one surface and hydrophobic polymer the opposite side can break oil-in-water selectively separate coalesced oil. This paper reports filter possesses additional function being antimicrobial. prepared by aero-spraying cotton fabric with solution copolymer containing poly(dimethylsiloxane) (PDMS) soaking other in dimethylaminoethyl methacrylate (DMAEMA). The then baked an oven to...
Graph representation plays an important role in the field of financial risk control, where relationship among users can be constructed a graph manner. In practical scenarios, relationships between nodes control tasks are bidirectional, e.g., merchants having both revenue and expense behaviors. neural networks designed for undirected graphs usually aggregate discriminative node or edge representations with attention strategy, but cannot fully exploit out-degree information when used built on...
Classifying an entire Whole Slide Image (WSI) in a single forward pass is challenging due to its vast resolution. Consequently, current effort on WSI classification resorts multiple instance learning (MIL), using patch-wise instances predict categories under image-wise supervision. However, recent MIL methods usually follow implicit selection strategy and ignore the effect from inherent patch category imbalances. In statistical sense, negative patches dominate WSIs provide sufficient samples...
Image retrieval aims to identify visually similar images within a database using given query image. Traditional methods typically employ both global and local features extracted from for matching, may also apply re-ranking techniques enhance accuracy. However, these often fail account the noise present in images, which can stem natural or human-induced factors, thereby negatively impacting performance. To mitigate this issue, we introduce novel setting low-quality image retrieval, propose an...
The performance of anomaly inspection in industrial manufacturing is constrained by the scarcity data. To overcome this challenge, researchers have started employing generation approaches to augment dataset. However, existing methods suffer from limited diversity generated anomalies and struggle achieve a seamless blending with original image. In paper, we these challenges new perspective, simultaneously generating pair overall image corresponding part. We propose DualAnoDiff, novel...
Human-human motion generation is essential for understanding humans as social beings. Although several transformer-based methods have been proposed, they typically model each individual separately and overlook the causal relationships in temporal sequences. Furthermore, attention mechanism transformers exhibits quadratic computational complexity, significantly reducing their efficiency when processing long In this paper, we introduce TIM (Temporal Interactive Modeling), an efficient...
Recent advancements in the field of Diffusion Transformers have substantially improved generation high-quality 2D images, 3D videos, and shapes. However, effectiveness Transformer architecture domain co-speech gesture remains relatively unexplored, as prior methodologies predominantly employed Convolutional Neural Network (CNNs) or simple a few transformer layers. In an attempt to bridge this research gap, we introduce novel Masked for generation, referred MDT-A2G, which directly implements...
Recently, methods like Zero-1-2-3 have focused on single-view based 3D reconstruction and achieved remarkable success. However, their predictions for unseen areas heavily rely the inductive bias of large-scale pretrained diffusion models. Although subsequent work, such as DreamComposer, attempts to make more controllable by incorporating additional views, results remain unrealistic due feature entanglement in vanilla latent space, including factors lighting, material, structure. To address...
Energy Internet is an important direction of energy development at the present stage. Based on research status home and abroad, this paper reviews architecture designs energy-information-value unified architecture. First all, layer, information layer value module, planning object, content, implementation methods are described; Then, interaction each module within between layers sorted out to form a logical relationship action mechanism three layers. Finally, considering influence current...
Comprehensive energy system planning is the primary core issue of Internet construction. The existing station location, capacity determination and pipeline network layout multi are generally obtained by staged optimization or traditional intelligent methods, resulting in final result non-optimal solution local optimal solution; At same time, only considers constraint cost, but does not consider later operation maintenance cost reduced construction comprehensive corridor, which will directly...
Globular clusters harbor numerous millisecond pulsars; however, the detection of long-period pulsars within these has been notably scarce. The search for encounters significant challenges due to pronounced red noise interference, necessitating crucial step removal in data preprocessing. In this study, we use running median filtering mitigate multiple globular cluster datasets obtained through observations with Five-hundred-meter Aperture Spherical radio Telescope (FAST). Additionally,...
With the development of times, a single form energy supply has been difficult to meet current load demand. The concept Internet came into being. It forms internet through various coupling equipment, such as electricity, gas and heat, so realize efficient utilization energy. In order maximize benefits, it is often necessary optimize capacity allocation reasonably allocate heterogeneous input. Considering equipment tolerance, set operation boundary. At present, most studies focus on analyzing...
Based on the development and status of solid state transformers, a class double-active clamping single-stage anti-excitation AC-AC converter is proposed. The circuit structure designed. Buck-Boost topology. Through leakage inductance capacitance resonance, purpose achieved. Relying high-frequency switching tubes zero-voltage to realize energy transformation. parameters storage transformer, output filter voltage current stress device are designed calculated in detail. main dual active clamped...