A5000699866- Flood Risk Assessment and Management
- Land Use and Ecosystem Services
- Adversarial Robustness in Machine Learning
- Hydrology and Watershed Management Studies
- Advanced Image and Video Retrieval Techniques
- Remote Sensing in Agriculture
- Advanced Neural Network Applications
- Environmental Changes in China
- Image Processing Techniques and Applications
- Remote Sensing and Land Use
- Robot Manipulation and Learning
- Recommender Systems and Techniques
- Human Pose and Action Recognition
- Anomaly Detection Techniques and Applications
- Marine and coastal ecosystems
- Digital Media Forensic Detection
- Advanced Differential Equations and Dynamical Systems
- Advanced Steganography and Watermarking Techniques
- Domain Adaptation and Few-Shot Learning
- Coastal and Marine Management
- Robotic Path Planning Algorithms
- Meromorphic and Entire Functions
- Hand Gesture Recognition Systems
- Laser Design and Applications
- Image Retrieval and Classification Techniques
Guangzhou Institute of Geography
2021-2024
Guangdong Academy of Sciences
2021-2024
Lanzhou Jiaotong University
2024
Jilin University
2024
Jilin Medical University
2024
Shenyang Institute of Automation
2004-2023
Chinese Academy of Sciences
2004-2023
Anhui Polytechnic University
2023
Southern Marine Science and Engineering Guangdong Laboratory (Guangzhou)
2022-2023
China Electronics Standardization Institute
2018-2023
Theano is a Python library that allows to define, optimize, and evaluate mathematical expressions involving multi-dimensional arrays efficiently. Since its introduction, it has been one of the most used CPU GPU compilers - especially in machine learning community shown steady performance improvements. being actively continuously developed since 2008, multiple frameworks have built on top produce many state-of-the-art models. The present article structured as follows. Section I provides an...
The development of object detection in the era deep learning, from R-CNN [11], Fast/Faster [10, 31] to recent Mask [14] and RetinaNet [24], mainly come novel network, new framework, or loss design. However, mini-batch size, a key factor for training neural networks, has not been well studied detection. In this paper, we propose Large Mini-Batch Object Detector (MegDet) enable with large size up 256, so that can effectively utilize at most 128 GPUs significantly shorten time. Technically,...
Traditional steganography methods often hide secret data by establishing a mapping relationship between and cover image or directly in noisy area, but has low embedding capacity. Based on the thought of deep learning, this paper, we propose new scheme based U-Net structure. First, form paired training, trained neural network includes hiding an extraction network; then, sender uses to embed into another full-size without any modification sends it receiver. Finally, receiver reconstruct...
Abstract Accurate water extraction and quantitative estimation of quality are two key challenging issues for remote sensing environment. Recent advances in big data, cloud computing, machine learning have promoted these fields into a new era. This study reviews the operating framework methods data environment monitoring, with emphasis on quality. The following aspects were investigated this study: (a) image source model evaluation metrics; (b) state‐of‐the‐art extraction, including...
Abstract Understanding the change in intensity and frequency of extreme precipitation plays an important role flood risk mitigation water resource management China. In this study, we analyzed abrupt changes long‐term trends over China from 1960 to 2015 based on daily stations. The possible teleconnection with large‐scale climate index was also been investigated. major results are as follows: (1) 14.72% 23.51% all stations have a point frequency, respectively. Moreover, most points occurred...
Urban lakes play an important role in urban development and environmental protection for the Wuhan agglomeration. Under impacts of urbanization climate change, understanding lake-water extent dynamics is significant. However, few studies on changes agglomeration exist. This research employed 1375 seasonally continuous Landsat TM/ETM+/OLI data scenes to evaluate from 1987 2015. The random forest model was used extract water bodies based eleven feature variables, including six remote-sensing...
The Chinese digital platform giants – Baidu, Alibaba and Tencent have quickly risen to be amongst the most notable developers users of artificial intelligence. One important catalyst for this development has been so-called Platform Business Group (PBG) strategy used by firms. In a firm aims develop powerful synergies tightly linking together number different platforms it owns so as offer multiple services under its umbrella. By applying PBG strategy, Alibaba, are able exploit enormous...
The use of remote sensing to monitor surface water bodies has gradually matured. Long-term serial change analysis and floods monitoring are currently research hotspots hydrology. However, these studies also faced with some problems, such as coarse temporal or spatial resolution data. In general, flood requires high resolution, small-scale extraction resolution. machine learning method been proven be effective against long-term extraction, random forests (RFs). MODIS data well suited for...
Abstract Multilayer feedforward neural networks (FFNs) are key in many machine learning models. They can be used to study firing rates, including synchronization and vibrational resonance (VR), under Gaussian white noise high-frequency stimulation (HFS). This examined how HFS affect synchronized rates VR different types of Izhikevich FFNs. In a ten-layer excitatory neuron network, increased from low high. Changes intensity, synaptic weights, time constants affected the propagation rates....
The core-collapse supernova remnant (SNR) Cassiopeia A (Cas A) is one of the brightest galactic radio sources with an angular radius $\sim$ 2.5 $\arcmin$. Although no extension this source has been detected in $\gamma$-ray band, using more than 1000 days LHAASO data above $\sim 0.8$ TeV, we find that its spectrum significantly softer those obtained Imaging Air Cherenkov Telescopes (IACTs) and flux near 1$ TeV about two times higher. In combination analyses 16 years \textit{Fermi}-LAT...
Depletion of water resources has threatened security in the Beijing-Tianjin-Hebei urban agglomeration, China. However, relative importance precipitation and urbanization to storage change not been sufficiently studied. In this study, both terrestrial (TWS) groundwater (GWS) Jing-Jin-Ji from 1979 2010s were investigated, based on global land data assimilation system (GLDAS) EartH2Observe (E2O) outputs, we used a night light index as an urbanization. The results showed that TWS anomaly varied...
Lakes have an important role in human life and the ecological environment, but they are easily affected by activity climate change, especially around urban areas. Hence, it is critical to extract water with a high precision method monitor long-term sequence dynamic changes lakes. As greatest natural lake of Beijing-Tianjin-Hebei region, Baiyangdian Lake has significant function life, socio-economic development, regional balance. This area shown large due change. The change monitoring process...
The accurate estimation of leaf chlorophyll content (LCC) is a significant foundation in assessing litchi photosynthetic activity and possible nutrient status. Hyperspectral remote sensing data have been widely used agricultural quantitative monitoring research for the non-destructive assessment LCC. Variable selection approaches are crucial analyzing high-dimensional datasets due to high danger overfitting, time-intensiveness, or substantial computational requirements. In this study,...
This study presents a comprehensive review of the primary distribution design an advanced network control system, emphasizing its evolution from initial requirements to practical applications. The system solves complex problems power management by combining real-time data analysis, intelligent decision making for resource allocation, rapid fault correction, remote monitoring and optimization methods, all aimed at ensuring stable safe operation grid. Its performance is geared towards fast...
The improvements in recent CNN-based object detection works, from R-CNN [11], Fast/Faster [10, 31] to Mask [14] and RetinaNet [24], mainly come new network, framework, or novel loss design. But mini-batch size, a key factor the training, has not been well studied. In this paper, we propose Large MiniBatch Object Detector (MegDet) enable training with much larger size than before (e.g. 16 256), so that can effectively utilize multiple GPUs (up 128 our experiments) significantly shorten time....
Concerned with the reliability of neural networks, researchers have developed verification techniques to prove their robustness. Most verifiers work real-valued networks. Unfortunately, exact (complete and sound) face scalability challenges provide no correctness guarantees due floating point errors. We argue that Binarized Neural Networks (BNNs) comparable robustness allow significantly more efficient verification. present a new system, EEV, for BNNs. EEV consists two parts: (i) novel SAT...
The catastrophic 8.0 Richter magnitude earthquake that occurred on 12 May 2008 in Wenchuan, China caused extensive damage to vegetation due widespread landslides and debris flows. In the past five years, Chinese government has implemented a series of measures restore severely afflicted area. How is recovering? It necessary important evaluate recovery effect earthquake-stricken areas. Based MODIS NDVI data from 2005 2013, area was extracted by quantified threshold detection method. rate after...