A5100734678- Reinforcement Learning in Robotics
- Advanced Neural Network Applications
- Adversarial Robustness in Machine Learning
- Advanced Image and Video Retrieval Techniques
- Machine Learning and Data Classification
- Face and Expression Recognition
- Music and Audio Processing
- Speech Recognition and Synthesis
- Domain Adaptation and Few-Shot Learning
- Evolutionary Algorithms and Applications
- Video Surveillance and Tracking Methods
- Recommender Systems and Techniques
- Advanced Graph Neural Networks
- Speech and Audio Processing
- Neural Networks and Applications
- Generative Adversarial Networks and Image Synthesis
- Network Security and Intrusion Detection
- Wheat and Barley Genetics and Pathology
- Advanced Multi-Objective Optimization Algorithms
- Advanced Algorithms and Applications
- Brain Tumor Detection and Classification
- Advanced MRI Techniques and Applications
- Image Retrieval and Classification Techniques
- Text and Document Classification Technologies
- Model Reduction and Neural Networks
Xianning Central Hospital
2025
Chinese Academy of Agricultural Sciences
2021-2024
Beijing University of Civil Engineering and Architecture
2018-2024
Chinese Academy of Sciences
2017-2024
Communication University of China
2023-2024
Northeastern University
2022-2024
Institute of Crop Sciences
2021-2024
Shandong Jiaotong University
2023-2024
University of Tübingen
2023
Xiamen University of Technology
2023
The concept of autoencoder was originally proposed by LeCun in 1987, early works on were used for dimensionality reduction or feature learning. Recently, with the popularity deep learning research, has been brought to forefront generative modeling. Many variants have different researchers and successfully applied many fields, such as computer vision, speech recognition natural language processing. In this paper, we present a comprehensive survey its various variants. Furthermore, also...
A vehicle detection algorithm is an indispensable component of intelligent traffic management and control systems, influencing the efficiency functionality system. In this paper, we propose a lightweight improvement method for YOLOv5 based on integrated perceptual attention, with few parameters high accuracy. First, module IPA Transformer encoder which leads to reduction in number while capturing global dependencies richer contextual information. Second, efficient multiscale spatial channel...
"Xinyi" (Magnolia biondii Pampan., Magnolia denudata Desr., and sprengeri Pampan.) is a traditional Chinese medicine listed in the Pharmacopoeia of People's Republic China. has anti-inflammatory, anti-allergic, antioxidant, antibacterial effects, significant curative effects treatment acute chronic rhinitis, allergic other rhinitis symptoms. However, comparative quality assessments these three species are scarce. This study investigated variations volatile oil content composition flower buds...
The overestimation phenomenon caused by function approximation is a well-known issue in value-based reinforcement learning algorithms such as deep Q-networks and DDPG, which could lead to suboptimal policies. To address this issue, TD3 takes the minimum value between pair of critics, introduces underestimation bias. By unifying these two opposites, we propose novel Weighted Delayed Deep Deterministic Policy Gradient algorithm, can reduce estimation error further improve performance weighting...
Abstract The non-coding RNA secondary structure largely determines its function. Hence, accuracy in acquisition is of great importance. Currently, this primarily relies on various computational methods. prediction the structures long sequences with high precision and reasonable cost remains challenging. Here, we propose a deep learning model, RNA-par, which could partition an sequence into several independent fragments (i-fragments) based exterior loops. Each i-fragment predicted...
Mango anthracnose is among the most severe diseases impacting mango yields and quality. While this disease can be effectively controlled through chemical means, it vital that appropriate field efficacy fate determination studies conducted when applying pesticides to crops in order appropriately gauge ecological health risks associated with use of these agents.GAP trials were explore efficacy, dissipation, terminal residues application mefentrifluconazole pyraclostrobin six locations...
Deep reinforcement learning (DRL) gives the promise that an agent learns good policy from high-dimensional information, whereas representation removes irrelevant and redundant information retains pertinent information. In this work, we demonstrate learned of Q-network its target should, in theory, satisfy a favorable distinguishable property. Specifically, there exists upper bound on similarity value functions two adjacent time steps typical DRL setting. However, through illustrative...
With the advancement of technology and development intelligence, self-balancing scooters have gradually developed in field short-distance travel become a fashionable, popular leisure means transportation. In addition, both export their safety are closely related to core patents. Therefore, order promote healthy automobile industry, this paper carries out following research: First, we introduce background patent research status value evaluation at home abroad. Then, considering fuzziness...
In speech recognition systems, it is a fundamental task to detect endpoints of signals. The efficiency endpoint detection affects the performance entire system. This especially important an isolated word Ordinarily, rate zero crossing and short time energy are used for detection. However, this not trivial process. fact, there many details we need pay attention to. Through experiments found that by selecting moderate algorithm ZCR energy, cooperating with preemphasis filter (prefiltering),...
Tibetan weedy barleys reside at the edges of qingke (hulless barley) fields in Tibet (Xizang). The spikes these contain or lack a brittle rachis, with either two- six-rowed and hulled hulless grains maturity. Although rachis trait is similar to that wild barley (Hordeum vulgare ssp. spontaneum Thell.), plants share genetic similarity domesticated barley. origin continues be debated. Here, we show most originated from cross-pollinated hybridization barleys, followed by hybrid self-pollination...
Small object detection is one of the research difficulties in detection, and Feature Pyramid Networks (FPN) a common feature extractor deep learning; thus, improving results small based on FPN great significance this field. In paper, SV-FPN proposed for task, which consists Object Enhancement (SOFE) Variance-guided Region Interest Fusion (VRoIF). When using as extractor, an SOFE module designed to enhance finer-resolution level maps from features are extracted. VRoIF takes variance RoI data...
Abstract Objective. To develop and evaluate a multi-path synergic fusion (MSF) deep neural network model for breast mass classification using digital tomosynthesis (DBT). Methods. We retrospectively collected 441 patients who had undergone DBT in which the regions of interest (ROIs) covering malignant/benign were extracted training validation. In proposed MSF framework, three multifaceted representations (gross mass, overview, background) are from ROIs independently processed by multi-scale...
During the training of a reinforcement learning (RL) agent, distribution data is non-stationary as agent's behavior changes over time. Therefore, there risk that agent overspecialized to particular and its performance suffers in larger picture. Ensemble RL can mitigate this issue by robust policy. However, it from heavy computational resource consumption due newly introduced value policy functions. In paper, avoid notorious resources issue, we design novel simple ensemble deep framework...
How to increase the connectivity of whole network is a critical research issue while achieving full coverage in WSNs. In this paper, novel paradigm regular topology proposed which adapts Cooperative Communication (CC) model with two-node cooperation achieve K-connectivity and coverage. Then we analyze changes sensor square triangle hexagon patterns transmission power under one-node operation respectively. Numerical results show our on could effectively, for