检索结果-内蒙古大学图书馆

IEEE Transactions on Artificial Intelligence 2024年第7期5卷 3690-3702页

作者： Xiao, Jiang-Wen Fang, Hongliang Wang, Yan-Wu Huazhong University of Science and Technology Ministry of Education School of Artificial Intelligence and Automation The Key Laboratory of Image Processing and Intelligent Control Wuhan430074 China

Short-term residential load forecasting is essential to demand side response. However, the frequent spikes in the load and the volatile daily load patterns make it difficult to accurately forecast the load. To deal with these problems, this article proposes a smoothing clustering method for daily load clustering and a pooling-ensemble model for one day ahead load forecasting. The whole short-term load forecasting framework in this article contains three steps. Specifically and first, the states of the residents are obtained by clustering the daily load curves with the proposed smoothing clustering method. Second, a weighted mixed Markov model is built to predict the probability distribution of the load state in the next day. Third, multiple predictors in the pooling-ensemble model are selected for different states and the load is forecasted by weighing the results of the multiple predictors based on the predicted states. Results of the case studies and comparison studies on two public datasets verify the advantages of the smoothing clustering method and the pooling-ensemble model. © 2020 IEEE.

关键词： Probability distributions

来源：评论

学校读者我要写书评

暂无评论

Improved 3-D LSTM: A Video Prediction Approach to Long Sequence Load Forecasting

引用

IEEE TRANSACTIONS ON SMART GRID 2025年第2期16卷 1885-1896页

作者： Xiao, Jiang-Wen Cui, Xue-Ying Liu, Xiao-Kang Fang, Hongliang Li, Peng-Cheng Huazhong Univ Sci & Technol Sch Artificial Intelligence & Automat Key Lab Image Proc & Intelligent Control Wuhan 430074 Peoples R China

Power load forecasting is the foundation of maintaining power grid stability, and can assist in decision-making to reduce operating costs. Fine-grained long sequence load forecasting contributes to formulating plans for power purchase, electricity consumption, energy storage, etc. Long sequence load forecasting requires models to effectively store memory and to accurately capture the long-term complex mapping between output and input. Therefore, this paper converts load sequences into three-dimensional (3D) continuous video frames and presents a model based on long short-term memory (LSTM) named the Improved 3D LSTM (I3D-LSTM) for predicting video frames. It contains two 3D LSTM units: For highly periodic load data, a Long-memory 3D LSTM unit is proposed, which has stronger long-term memory and removes short-term memory;On weakly periodic datasets, a Simplified 3D LSTM unit without the scoring parts exhibits excellent performance. I3D-LSTM also contains a 3D recurrent neural network architecture with residual. Dropblock and batch normalization are integrated into the I3D-LSTM, which are analyzed as excellent solutions for overfitting in 3D LSTM. Comprehensive tests are conducted on different sequence lengths in multiple real-world datasets. Comparison results indicate that I3D-LSTM outperforms various advanced models.

关键词： Long short term memory Three-dimensional displays Load modeling Forecasting Load forecasting Convolution Time series analysis Long sequence time series forecasting load forecasting long short-term memory deep learning

来源：评论

学校读者我要写书评

暂无评论

Differentially Private Linearized ADMM Algorithm for Decentralized Nonconvex Optimization

引用

IEEE TRANSACTIONS ON INFORMATION FORENSICS AND SECURITY 2025年 20卷 3316-3329页

作者： Yue, Xiao-Yu Xiao, Jiang-Wen Liu, Xiao-Kang Wang, Yan-Wu Huazhong Univ Sci & Technol Sch Artificial Intelligence & Automat Key Lab Image Proc & Intelligent Control Wuhan 430074 Peoples R China

Privacy preservation is a challenging problem in decentralized nonconvex optimization containing sensitive data. Prior approaches to decentralized nonconvex optimization are either not strong enough to protect privacy or exhibit low utility under a high privacy guarantee. To address these issues, we propose a differentially private linearized alternating direction method of multipliers (DP-LADMM), which achieves fast convergence property for nonconvex objective functions while achieving saddle/maximum avoidance under differential privacy guarantee. We also apply the Analytic Gaussian Mechanism to track the cumulative privacy loss and provide a tight global differential privacy guarantee for DP-LADMM. The theoretical analysis offers an explicit convergence rate for our algorithm. To the best of our knowledge, this is the first paper to provide explicit convergence for decentralized nonconvex optimization with differential privacy and saddle/maximum avoidance. Numerical simulations and comparison studies on decentralized estimation confirm the superiority of the algorithm and the effectiveness of global privacy preservation.

关键词： Optimization Privacy Differential privacy Convergence Linear programming Vectors Machine learning Training Protection Perturbation methods Decentralized nonconvex optimization linearized alternating direction method of multipliers privacy preservation saddle avoidance

来源：评论

学校读者我要写书评

暂无评论

Spatial cascaded clustering and weighted memory for unsupervised person re-identification

引用

image AND VISION COMPUTING 2025年 156卷

作者： Hong, Jiahao Zuo, Jialong Han, Chuchu Zheng, Ruochen Tian, Ming Gao, Changxin Sang, Nong Huazhong Univ Sci & Technol Sch Artificial Intelligence & Automat Key Lab Image Proc & Intelligent Control Wuhan Hubei Peoples R China

Recent advancements in unsupervised person re-identification (re-ID) methods have demonstrated high performance by leveraging fine-grained local context, often referred to as part-based methods. However, many existing part-based methods rely on horizontal division to obtain local contexts, leading to misalignment issues caused by various human poses. Moreover, misalignment of semantic information within part features hampers the effectiveness of metric learning, thereby limiting the potential of part-based methods. These challenges result in under-utilization of part features in existing approaches. To address these issues, we introduce the Spatial Cascaded Clustering and Weighted Memory (SCWM) method. SCWM aims to parse and align more accurate local contexts for different human body parts while allowing the memory module to balance hard example mining and noise suppression. Specifically, we first analyze the issues of foreground omissions and spatial confusions in previous methods. We then propose foreground and space corrections to enhance the completeness and reasonableness of human parsing results. Next, we introduce a weighted memory and utilize two weighting strategies. These strategies address hard sample mining for global features and enhance noise resistance for part features, enabling better utilization of both global and part features. Extensive experiments conducted on Market-1501, DukeMTMC-reID and MSMT17 datasets validate the effectiveness of the proposed method over numerous state-of-the-art methods.

关键词： Person re-identification Fine-grained Unsupervised learning Contrastive learning

来源：评论

学校读者我要写书评

暂无评论

Event-Triggered H∞ control for Fuzzy Two-Time-Scale Systems

引用

IEEE TRANSACTIONS ON FUZZY SYSTEMS 2024年第4期32卷 2211-2222页

作者： Yu, Tian-Tian Wang, Yan-Wu Lei, Yan Liu, Xiao-Kang Huazhong Univ Sci & Technol Sch Artificial Intelligence & Automat Key Lab Image Proc & Intelligent Control Wuhan 430074 Peoples R China

This article focuses on the H-infinity control for nonlinear two-time-scale systems with event-triggered mechanisms. Utilizing the Takagi-Sugeno fuzzy model, it is feasible to represent nonlinear two-time-scale systems as fuzzy two-time-scale systems. Event-triggered state feedback control strategy is designed for achieving the H-infinity performance, which inevitably leads to the asynchronous phenomenon of the premise variables between continuous time and triggering instants. Under the consideration of the asynchronous phenomenon, based on a H-infinity-dependent Lyapunov function, the fuzzy composite state feedback controller gains and the event-triggered parameters are codesigned in the form of linear matrix inequalities, and the upper bound of H-infinity is provided as well. Furthermore, the proposed event-triggered mechanism ensures the exclusion of Zeno behavior. Finally, simulation results including comparison studies are shown to demonstrate the effectiveness of the proposed control strategy.

关键词： Asynchronous phenomenon event-triggered H-infinity control fuzzy two-time-scale system LMI

来源：评论

学校读者我要写书评

暂无评论

Color and Luminance Separated Enhancement for Low-Light images with Brightness Guidance

引用

SENSORS 2024年第9期24卷 2711页

作者： Zhang, Feng Liu, Xinran Gao, Changxin Sang, Nong Huazhong Univ Sci & Technol Sch Artificial Intelligence & Automat Key Lab Image Proc & Intelligent Control Wuhan 430074 Peoples R China

Existing retinex-based low-light image enhancement strategies focus heavily on crafting complex networks for Retinex decomposition but often result in imprecise estimations. To overcome the limitations of previous methods, we introduce a straightforward yet effective strategy for Retinex decomposition, dividing images into colormaps and graymaps as new estimations for reflectance and illumination maps. The enhancement of these maps is separately conducted using a diffusion model for improved restoration. Furthermore, we address the dual challenge of perturbation removal and brightness adjustment in illumination maps by incorporating brightness guidance. This guidance aids in precisely adjusting the brightness while eliminating disturbances, ensuring a more effective enhancement process. Extensive quantitative and qualitative experimental analyses demonstrate that our proposed method improves the performance by approximately 4.4% on the LOL dataset compared to other state-of-the-art diffusion-based methods, while also validating the model's generalizability across multiple real-world datasets.

关键词： low-light image enhancement diffusion models Retinex decomposition strategy brightness guidance

来源：评论

学校读者我要写书评

暂无评论

Distributed Privacy-Preserving Secondary control for DC Microgrids via State Decomposition

引用

IEEE TRANSACTIONS ON SUSTAINABLE ENERGY 2024年第2期15卷 726-734页

作者： Yuan, Qi-Fan Wang, Yan-Wu Liu, Xiao-Kang Liu, Zhi-Wei Huazhong Univ Sci & Technol Sch Artificial Intelligence & Automat Key Lab Image Proc & Intelligent Control Wuhan 430074 Peoples R China

In this article, a distributed secondary controller is proposed for an islanded DC microgrid to preserve privacy information of the local power sources, while achieving current sharing among the generators and voltage regulation of the DC bus. Based on the state decomposition mechanism, the original secondary control signal of each generator is decomposed into two parts: one takes over the role of communicating with neighbors, while the other affects the evolution of the former one and is invisible to the neighbors. Besides, random noises are introduced to mask the true value of the transmitted signals. Under the proposed controller, the current data of the local generators can be kept confidential, as well as the associated privacy power information. In this way, the power security risks caused by privacy information leakage can be avoided. Lastly, both simulation and experiment tests are imposed to verify the effectiveness of the proposed control scheme.

关键词： Privacy Microgrids Voltage control Generators Power control Matrix converters Security DC microgird distributed secondary control privacy preservation state decomposition current sharing voltage regulation

来源：评论

学校读者我要写书评

暂无评论

FADE: A Task-Agnostic Upsampling Operator for Encoder-Decoder Architectures

引用

INTERNATIONAL JOURNAL OF COMPUTER VISION 2025年第1期133卷 151-172页

作者： Lu, Hao Liu, Wenze Fu, Hongtao Cao, Zhiguo Huazhong Univ Sci & Technol Sch Artificial Intelligence & Automat Key Lab Image Proc & Intelligent Control Minist Educ Wuhan 430074 Peoples R China

The goal of this work is to develop a task-agnostic feature upsampling operator for dense prediction where the operator is required to facilitate not only region-sensitive tasks like semantic segmentation but also detail-sensitive tasks such as image matting. Prior upsampling operators often can work well in either type of the tasks, but not both. We argue that task-agnostic upsampling should dynamically trade off between semantic preservation and detail delineation, instead of having a bias between the two properties. In this paper, we present FADE, a novel, plug-and-play, lightweight, and task-agnostic upsampling operator by fusing the assets of decoder and encoder features at three levels: (i) considering both the encoder and decoder feature in upsampling kernel generation;(ii) controlling the per-point contribution of the encoder/decoder feature in upsampling kernels with an efficient semi-shift convolutional operator;and (iii) enabling the selective pass of encoder features with a decoder-dependent gating mechanism for compensating details. To improve the practicality of FADE, we additionally study parameter- and memory-efficient implementations of semi-shift convolution. We analyze the upsampling behavior of FADE on toy data and show through large-scale experiments that FADE is task-agnostic with consistent performance improvement on a number of dense prediction tasks with little extra cost. For the first time, we demonstrate robust feature upsampling on both region- and detail-sensitive tasks successfully. Code is made available at: https://***/poppinace/fade

关键词： Feature upsampling Dense prediction Semantic segmentation image matting Object detection Instance segmentation Depth estimation

来源：评论

学校读者我要写书评

暂无评论

Distributed robust iterative learning control for AC microgrids with external disturbances

引用

SCIENCE CHINA-INFORMATION SCIENCES 2025年第10期68卷 1-21页

作者： Su, Hongyu Lai, Jingang Huazhong Univ Sci & Technol Sch Artificial Intelligence & Automat Key Lab Image Proc & Intelligent Control Minist Educ Wuhan 430074 Peoples R China

This study proposes a distributed secondary control scheme based on distributed robust iterative learning control (DRILC) for islanded alternating current (AC) microgrids subjected to external disturbances. By employing the decoupled sliding mode consensus approach, voltage regulation, frequency restoration, and accurate active power sharing can be achieved within a finite time on the proposed novel integral terminal sliding mode (ITSM) manifold. Furthermore, the appropriate iterative update law in the ITSM-based controller is utilized to learn and eliminate external disturbances as effectively as possible. In the proposed control scheme, the iterative learning control and sliding mode control are designed to function in a complementary manner, which enhances the performance of the secondary control scheme for multi-objective regulations. The stability criteria and robustness to external disturbances of the closed-loop microgrid system in the iteration and time domains are also rigorously derived with the help of the Lyapunov direct method. Finally, the effectiveness of the proposed secondary control scheme is validated by case studies of an islanded AC microgrid test system in the MATlab/SimPowerSystems software environment.

关键词： distributed secondary control islanded microgrid iterative learning control sliding mode control external disturbances

来源：评论

学校读者我要写书评

暂无评论

Enhancing Deep Learning-based Depression Level Estimation Based on Multi-task Learning 24

Enhancing Deep Learning-based Depression Level Estimation Ba...

引用

7th International Conference on Machine Learning and Machine Intelligence, MLMI 2024

作者： Ye, Qingxin Xie, Zhenming Sun, Hao Xin, Luyao Chen, Youwen Song, Jian Chen, Yen-Wei Provincial key Laboratory of Computational Science Huaqiao University Quanzhou China Intelligent Image Processing Lab Zhejiang University Hangzhou China Intelligent Image Processing Lab Ritsumeikan University Osaka Japan

ISBN: (纸本)9798400717833

In recent years, depression, as a serious mental illness, has received widespread attention from various sectors of society. How to identify depressive emotions in a timely manner and detect depression has become an urgent issue that needs to be addressed. Currently, many experts and scholars propose using deep learning methods to assess the level of depression. Compared to other deep learning models, Transformer models have parallel computing capabilities with their self-attention mechanism, enabling them to handle entire sequences and capture long-distance dependencies. After the success of Transformers in natural language processing, similar architectures have been introduced into the field of computer vision, promoting advancements in multi-modal research. However, how to integrate other information or tasks to improve assessment accuracy remains a challenge. In our team's preliminary work, we proposed a multi-modal adaptive fusion Transformer network for assessing depression levels. Inspired by multi-task learning, we introduced gender classification as a second auxiliary task alongside the main task of depression regression, forming a three-task learning framework including depression classification. This was done to enhance the model's performance. We validated the effectiveness of the proposed method by testing it on a public dataset (AVEC 2019 Detecting Depression with AI Sub-challenge). The experimental results show that our proposed method achieved a Concordance Correlation Coefficient (CCC) of 0.60448, which is 1.688% higher than the accuracy of the dual-task learning method (CCC = 0.5876). © 2024 Copyright held by the owner/author(s).

关键词： Multi-task learning

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：