A5048159038- Retinal Imaging and Analysis
- Medical Image Segmentation Techniques
- Emotion and Mood Recognition
- Reservoir Engineering and Simulation Methods
- Retinal Diseases and Treatments
- Hydraulic Fracturing and Reservoir Analysis
- Liver physiology and pathology
- Brain Tumor Detection and Classification
- Oil and Gas Production Techniques
- Structural Response to Dynamic Loads
- Mesenchymal stem cell research
- Dermatology and Skin Diseases
- Speech and Audio Processing
- COVID-19 diagnosis using AI
- Lattice Boltzmann Simulation Studies
- 3D Surveying and Cultural Heritage
- Caching and Content Delivery
- Hematopoietic Stem Cell Transplantation
- Speech Recognition and Synthesis
- Integrated Circuits and Semiconductor Failure Analysis
- Advanced Graph Neural Networks
- Advanced Image Processing Techniques
- Machine Learning in Materials Science
- Fluid Dynamics and Vibration Analysis
- Structural Behavior of Reinforced Concrete
Dian Diagnostics (China)
2020
BOE Technology Group (China)
2020
Inception Institute of Artificial Intelligence
2019
PetroChina Southwest Oil and Gas Field Company (China)
2008-2011
Research Institute of Petroleum Exploration and Development
2008-2011
Shanghai Stomatological Hospital
2005
Medical image analysis has two important research areas: disease grading and fine-grained lesion segmentation. Although the former problem often relies on latter, are usually studied separately. Disease severity can be treated as a classification problem, which only requires image-level annotations, while segmentation stronger pixel-level annotations. However, pixel-wise data annotation for medical images is highly time-consuming domain experts. In this paper, we propose collaborative...
Diabetic retinopathy (DR) is a complication of diabetes that severely affects eyes. It can be graded into five levels severity according to international protocol. However, optimizing grading model have strong generalizability requires large amount balanced training data, which difficult collect, particularly for the high levels. Typical data augmentation methods, including random flipping and rotation, cannot generate with diversity. In this paper, we propose diabetic generative adversarial...
This paper introduces a novel neural network - flow completion (FCN) to infer the fluid dynamics, includ-ing field and force acting on body, from incomplete data based Graph Convolution AttentionNetwork. The FCN is composed of several graph convolution layers spatial attention layers. It designed inferthe velocity vortex contribution when combined with map (VFM)method. Compared other networks adopted in capable dealing bothstructured unstructured data. performance proposed assessed by...
SimRank is a classic measure of the similarities nodes in graph. Given node u graph $G =(V, E)$, \em single-source query returns $s(u, v)$ between and each $v \in V$. This type queries has numerous applications web search social networks analysis, such as link prediction, mining, spam detection. Existing methods for queries, however, incur cost at least linear to number n, which renders them inapplicable real-time interactive analysis. paper proposes \prsim, an algorithm that exploits...
Speech emotion recognition (SER) has attracted great attention in recent years due to the high demand for emotionally intelligent speech interfaces. Deriving speaker-invariant representations is crucial. In this paper, we propose apply adversarial training SER learn representations. Our model consists of three parts: a representation learning sub-network with time-delay neural network (TDNN) and LSTM statistical pooling, an classification speaker network. Both take output as input. Two...
The morphological ECG features for arrhythmia diagnosis are usually identified and combined on different scales.For example, can be the scale of length or amplitude QRS waves. Professionals then make a based combination these features. Attention-based deep neural networks have been proved to boost meaningful scales suppress weak To combine classification, we proposed MADNN: multi-scale attention network classification. Our was designed combining kernel-wise branch-wise modules backbone...
CAPTCHAs based on reading text are susceptible to machine-learning-based attacks due recent significant advances in deep learning (DL). To address this, this paper promotes image/visual captioning CAPTCHAs, which is robust against attacks. develop image/visual-captioning-based proposes a new image architecture by exploiting tensor product representations (TPR), structured neural-symbolic framework developed cognitive science over the past 20 years, with aim of integrating DL explicit...