A5101475937- AI in cancer detection
- Traffic Prediction and Management Techniques
- Advanced Neural Network Applications
- Traffic control and management
- Advanced Image and Video Retrieval Techniques
- Image Enhancement Techniques
- Autonomous Vehicle Technology and Safety
- Digital Imaging for Blood Diseases
- Heat Transfer and Boiling Studies
- Ultrasound Imaging and Elastography
- Advanced Image Processing Techniques
- Image Retrieval and Classification Techniques
- Advanced Vision and Imaging
- Transportation Planning and Optimization
- Radiomics and Machine Learning in Medical Imaging
- Biomedical Text Mining and Ontologies
- Machine Learning in Healthcare
- Human Pose and Action Recognition
- Reinforcement Learning in Robotics
- Transportation and Mobility Innovations
- Face recognition and analysis
- Image and Signal Denoising Methods
- Flow Measurement and Analysis
- Image Processing Techniques and Applications
- Microfluidic and Capillary Electrophoresis Applications
Beijing University of Posts and Telecommunications
2025
Alibaba Group (United States)
2023
Shanghai Jiao Tong University
2020-2022
Shanghai Medical Information Center
2022
Federal Highway Administration
2015
Turner-Fairbank Highway Research Center
2011-2012
United States Department of Transportation
2012
Technical University of Denmark
1988-1992
There have been considerable debates over 2D and 3D representation learning on medical images. approaches could benefit from large-scale pretraining, whereas they are generally weak in capturing large contexts. natively strong contexts, however few publicly available dataset is diverse enough for universal pretraining. Even hybrid (2D + 3D) approaches, the intrinsic disadvantages within / parts still exist. In this study, we bridge gap between convolutions by reinventing convolutions. We...
The 3D Lookup Table (3D LUT) is a highly-efficient tool for real-time image enhancement tasks, which models non-linear color transform by sparsely sampling it into discretized lattice. Previous works have made efforts to learn image-adaptive output values of LUTs flexible but neglect the importance strategy. They adopt sub-optimal uniform point allocation, limiting expressiveness learned since (tri-)linear interpolation between points in LUT might fail model local non-linearities transform....
Multi-agent reinforcement learning (MARL) has proven to be effective and promising in team collaboration tasks. Knowledge transfer MARL also received increasing attention. Compared knowledge single-agent tasks, multi-agent tasks is more complex due the need account for coordination among agents. However, existing transfer-based methods only focus on strategies or agent-level a single task, of such specific task new different types likely fail. In this paper, we propose multitask-based...
When applying multi-instance learning (MIL) to make predictions for bags of instances, the prediction accuracy an instance often depends on not only itself but also its context in corresponding bag. From viewpoint causal inference, such bag contextual prior works as a confounder and may result model robustness interpretability issues. Focusing this problem, we propose novel interventional (IMIL) framework achieve deconfounded instance-level prediction. Unlike traditional likelihood-based...
Deep convolutional neural networks have achieved great progress in image denoising tasks. However, their complicated architectures and heavy computational cost hinder deployments on mobile devices. Some recent efforts designing lightweight focus reducing either FLOPs (floating-point operations) or the number of parameters. these metrics are not directly correlated with on-device latency. In this paper, we identify real bottlenecks that affect CNN-based models' runtime performance devices:...
Heterogeneous driver behavior during safety-critical events is more complicated than normal driving situations and difficult to capture by statistical models. This paper applies an agent-based reinforcement learning method represent heterogeneous for different drivers events. The naturalistic data of are used in agent training. As output the Neuro-Fuzzy Actor Critic Reinforcement Learning (NFACRL) training technique, rules embedded agents actions between drivers. results show that NFACRL...
Previous deep learning based Computer Aided Diagnosis (CAD) system treats multiple views of the same lesion as independent images. Since an ultrasound image only describes a partial 2D projection 3D lesion, such paradigm ignores semantic relationship between different which is inconsistent with traditional diagnosis where sonographers analyze from at least two views. In this paper, we propose multi-task framework that complements Benign/Malignant classification task recognition (LR) helps...
A transit-time flow meter, using periodic temperature fluctuations as tracers, has been developed for measuring liquid small 0.1 ml/min in microchannels. By injecting square waves of heat into the upstream with a tiny resistance wire heater, are generated downstream. The fundamental frequency phase shift signal respect to wave is found be linear function reciprocal mean velocity fluid. principle enables meter have high accuracy, better than 0.2%, and good linearity. This will used measure...
An agent-based multi-layer reinforcement learning (RL) framework for naturalistic driving behavior simulation in traffic is introduced. Each agent a replication of an individual driver. implemented by applying artificial intelligence concepts, including: fuzzy logic, neural networks, and algorithms. A revised Neuro-Fuzzy Actor Critic Reinforcement Learning (NFACRL) proposed to simulate vehicle actions during safety-critical events when the state complicated. The NFACRL algorithm can handle...
Infrastructure rehabilitation is critical to assuring the proper functioning of a nation’s transportation infrastructure. While goal typically restore or improve performance system, construction activity itself disrupts traffic and worsens already-congested network with unsafe work zones. Traffic simulation models are becoming widely used in evaluating such disruption controlled environment without physical implementation. Though rapid increase computer processing power has made microscopic...
When applying multi-instance learning (MIL) to make predictions for bags of instances, the prediction accuracy an instance often depends on not only itself but also its context in corresponding bag. From viewpoint causal inference, such bag contextual prior works as a confounder and may result model robustness interpretability issues. Focusing this problem, we propose novel interventional (IMIL) framework achieve deconfounded instance-level prediction. Unlike traditional likelihood-based...
An agent-based, artificial intelligence technique known as reinforcement learning has been used to capture vehicle behavior and simulate driver's actions in traffic, especially during emergency situations. This paper discusses the training parameters their influence on agent simulation performance. A type of called Neuro-Fuzzy Actor Critic Reinforcement Learning (NFACRL) is test with an objective improving systematic parameter determination optimization methodology provided.
Early diagnosis of signet ring cell carcinoma dramatically improves the survival rate patients. Due to lack public dataset and expert-level annotations, automatic detection on (SRC) has not been thoroughly investigated. In MICCAI DigestPath2019 challenge, apart from foreground (SRC region)-background (normal tissue area) class imbalance, SRCs are partially annotated due costly medical image annotation, which introduces extra label noise. To address issues simultaneously, we propose Decoupled...