A5090377229- Oceanographic and Atmospheric Processes
- Climate variability and models
- Ocean Waves and Remote Sensing
- Sentiment Analysis and Opinion Mining
- Tropical and Extratropical Cyclones Research
- Topic Modeling
- Meteorological Phenomena and Simulations
- Customer churn and segmentation
- Geological formations and processes
- Video Surveillance and Tracking Methods
- Advanced Sensor and Control Systems
- Remote Sensing and Land Use
- Geology and Paleoclimatology Research
- Speech and Audio Processing
- Text and Document Classification Technologies
- Advanced Decision-Making Techniques
- Coastal and Marine Dynamics
- Cardiac Imaging and Diagnostics
- Integrated Energy Systems Optimization
- Polymer-Based Agricultural Enhancements
- Music Technology and Sound Studies
- Electrical Fault Detection and Protection
- Thermochemical Biomass Conversion Processes
- Reservoir Engineering and Simulation Methods
- Face recognition and analysis
University of Shanghai for Science and Technology
2025
Ocean University of China
2008-2024
Dalian University of Technology
2009-2023
Northeast Agricultural University
2023
Alibaba Group (China)
2023
Texas A&M University
2012-2022
Tianjin University
2014-2022
Qingdao National Laboratory for Marine Science and Technology
2021
Sinopec (China)
2021
Deepblue Technology (China)
2020
Improving a tropical cyclone's forecast and mitigating its destructive potential requires knowledge of various environmental factors that influence the path intensity. Herein, using combination observations model simulations, we systematically demonstrate cyclone intensification is significantly affected by salinity-induced barrier layers, which are "quasi-permanent" features in upper oceans. When cyclones pass over regions with increased stratification stability within layer reduce...
Abstract Tropical Cyclones (TCs) are devastating natural disasters. Analyzing four decades of global TC data, here we find that among all TC-active basins, the South China Sea (SCS) stands out as particularly difficult ocean for TCs to intensify, despite favorable atmosphere and conditions. Over SCS, intensification rate its probability a rapid (intensification by ≥ 15.4 m s −1 day ) only 1/2 1/3, respectively, those rest world ocean. Originating from complex interplays between astronomic...
Excessive drinking has been listed by the World Health Organization as fifth major risk factor; especially liver, core organ of alcohol metabolism, is prone to organic lesions. Probiotics have received attention due their bioactivity for liver protection. The beneficial effects probiotics on hosts are related physiological functions. Therefore, based concept second-generation synbiotes, this study explored protective four dietary polyphenols stress tolerance, hydrophobicity, adhesion, and...
The Bel Canto performance is a complex and multidimensional art form encompassing pitch, timbre, technique, affective expression. To accurately reflect performer's singing proficiency, it essential to quantify evaluate their vocal technical execution precisely. Convolutional Neural Networks (CNNs), renowned for robust ability capture spatial hierarchical information, have been widely adopted in various tasks, including audio pattern recognition. However, existing CNNs exhibit limitations...
Model high-protein nutrition bars (HPNBs) were formulated by incorporating whey protein isolate (WPI) and casein (CN) at various extrusion temperatures (50, 75, 100, 125, 150 °C) with a content of 45 g per 100 g. The free sulfhydryl groups, amino hardness, microstructures HPNBs analyzed periodically 37 °C over storage period days. Specifically, group, group surface hydrophobicity extruded (WPE) (CE) significantly reduced (P < 0.05) compared to those unextruded protein. WPE (HWPE) CE (HWCE)...
Conversion rate (CVR) prediction is an essential task for e-commerce platforms. However, refunds frequently occur after conversion in online shopping systems, which drives us to pay attention effective building healthier services. This paper defines the probability of item purchasing without any subsequent refund as (ECVR). A simple paradigm ECVR decompose it into two sub-tasks: CVR and post-conversion (RFR) prediction. RFR suffers from data sparsity (DS) sample selection bias (SSB) issues,...
Aiming at the problem that traditional convolutional neural networks cannot fully capture text features during feature extraction, and a single model effectively extract deep features, this paper proposes sentiment classification method based on attention mechanism of LSTM-TCN hybrid model. First, use Word2vec word vector to transform words into form vectors. Secondly, with Long Short-Term Memory (LSTM) obtain serialized information text, then combined Temporal Convolutional Network (TCN)...
Abstract This paper studies and implements a face recognition system based on convolutional neural network. The firstly uses AdaBoost algorithm to detect the face, that is get position size of accurately in image, then deep convolution network extract features classify them. Finally, hardware application Altera DE1_SOC development board, designs completes with high rate.
In view of the simple structure a single neural network model, traditional convolutional cannot fully extract deep text features. This paper proposes sentiment classification based on Attention Mechanism and Decomposition Convolutional Neural Network model. First, comprehensive features are obtained with help parallel (DCNN). Secondly, it combines attention mechanism to important feature information improve optimized effect. Finally, is sentimentally classified at layer. After conducting...
The industry of biomass power in China has been growing tremendously recent years, but it is faced with many challenges now. low efficiency straw combustion the boiler results that carbon content fly ash could reach 15~30%. also rich silica, potassium, calcium and a lot microelements. A kind from plant mainly uses corn as feedstock was studied. XRF analysis showed contents K 2 O, CaO MgO were 7.22%, 13.61% respectively. Leaching experiments conducted to evaluate solubility nutritive elements...
The task of person re-identification (re-id) is to match images people observed in different camera views. Recent researches mainly focus on feature representation and metric learning. Many global learning approaches have achieved good performance. Since comparing all the samples with a single inappropriate handle heterogeneous data, some local are proposed. But most them cannot be used re-id directly due research challenges. Also, they usually need complicated computation solve optimization...
In this paper, an optimal scheduling method of cold-heat-electricity combined power supply microgrid for wind accommodation is proposed. Firstly, the principle heat pump (HP) and integrated electric heating demand response to promote was analyzed, source-charge cooperative operation strategy multi-energy micro grid constructed. The two-layer model adopted, upper layer adopts price-type increase off-load consumption capacity. lower takes cost minimization as goal, adds load role HP,...
EEG feature identification and body's sensing are used together to confirm that the current flowing through human body reaches perception threshold, since existing test method is sensitive subject's subjective awareness experimental data's reliability consistency not good. Based on method, relationship studied among effect threshold many influencing factors such as touch voltage, area, pressure, skin type, environmental temperature humidity, time period in which passes body, frequency,...
As One of Features from other Languages, the Chinese Tone Changes are Mainly Decided by its Vowels, so Vowel Variation Becomes Important in Speech Recognition Research. Normal Ways Always Based on Fundamental Frequency Signal, which can Not Keep Integrity Signal. we Bring Forward to a Mathematical Morphological Processing Spectrograms for Vowels. Firstly, will have Pretreatment Recording Good Signal Using Cooledit Pro Software, and Converted into Spectrograms; Secondly, do Smooth Normalized...
Abstract The development of high-resolution, fully coupled, regional Earth system model systems is important for improving our understanding climate variability, future projections, and extreme events at scales. Here we introduce present an overview the newly developed Regional Community System Model (R-CESM). Different from other existing models, R-CESM based on version 2 (CESM2) framework. We have incorporated Weather Research Forecasting (WRF) Ocean Modeling (ROMS) into CESM2 as...