检索结果-内蒙古大学图书馆

23rd China National Conference on Computational Linguistics, CCL 2024

作者： Liu, Xiao Li, Ying Yu, Zhengtao Faculty of Information Engineering and Automation Kunming University of Science and Technology Kunming China Yunnan Key Laboratory of Artificial Intelligence Kunming China

ISBN: (纸本)9789819783663

Pre-trained language model-based methods for Chinese Grammatical Error Correction (CGEC) are categorized into Seq2Seq and Seq2Edit types. However, both Seq2Seq and Seq2Edit models depend on high-quality training data significantly. Considering the strong generation and inference ability of large language models (LLMs), we propose a large language model-guided optimization training method to exploit LLMs to extract error knowledge to optimize the traditional CGEC model training process. On the one hand, we use error types and confusion sets as extra knowledge to guide LLMs to generate diverse pseudo data, thus extending the error distribution of our training data. On the other hand, LLMs are utilized to infer the predicted results from our CGEC models and obtain the re-training data, thus iteratively optimizing our pre-trained CGEC models. Experiments on two benchmark datasets show that our LLMs-guided optimization method with small-scale training data can achieve comparable results with baseline models with large-scale training data. Detailed comparison experiments demonstrate that both the early deviser pseudo data and the later re-training data are extremely useful for traditional CGEC model optimization training, and can benefit from each other. We will release our code and prompts at https://***/SakuraAcedia/llm-cgec-got to facilitate future work. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

关键词： Error correction

来源：评论

学校读者我要写书评

暂无评论

A Remaining Useful Life Prediction of Turbofan Engines Based on Multi-Scale Temporal Convolutional Networks with Dual Squeeze-Excitation Attention Mechanism

A Remaining Useful Life Prediction of Turbofan Engines Based...

引用

2024 International Conference on Computational Intelligence and Communication System, CCICS 2024

作者： Wang, Hairui Li, Dongjun Li, Dongwen Li, Ya Zhu, Guifu Faculty of Information Engineering and Automation Kunming University of Science and Technology Yunnan Kunming650504 China Information Technology Construction Management Center Kunming University of Science and Technology Yunnan Kunming650504 China

To better handle temporal data and delve into learning the features of the data, a turbofan engine residual life prediction method is proposed, which integrates a dual-squeeze-excitation attention mechanism with a multi-scale temporal convolutional network. Firstly, utilizing a sliding window, the extracted multi-dimensional sensor features undergo overlapping sampling to enhance the model's perception of temporal data. Secondly, a hybrid network prediction model based on DSE-MTCN is constructed, employing multi-scale convolutional kernels to expand the receptive field of convolution, assigning different weights to features, and adaptively allocating weights to hidden layer units. Lastly, the DSE-MTCN prediction model is globally optimized using the RAdam algorithm. The results demonstrate that this method effectively enhances the accuracy and generalization ability of the prediction model. © Published under licence by IOP Publishing Ltd.

关键词： Prediction models

来源：评论

学校读者我要写书评

暂无评论

Convolutional Neural Networks Aided Reinforcement Learning for Accelerated Optimization of Antenna Topology

引用

Applied Computational Electromagnetics Society Journal 2025年第1期40卷 35-41页

作者： Dou, Jiangling Gong, Hao Wei, Shuaibing Chen, Haokang Chen, Yujie Shen, Tao Song, Jian Yunnan Key Laboratory of Computer Technologies Application Kunming University of Science and Technology Kunming650500 China School of Information Engineering and Automation Kunming University of Science and Technology Kunming650500 China College of Mechanical and Electrical Engineering Yunnan Electromechanical Vocational and Technical College Kunming650500 China

A machine learning (ML) framework is proposed to achieve the automatic and rapid optimization of antenna topologies. A convolutional neural network (CNN) is utilized as a surrogate model (SM) and is combined with reinforcement learning (RL) algorithms. Specifically, the RL agent interacts with simulation software to learn. Data accumulated from electromagnetic (EM) simulations are used to train the SM. The CNN-based SM predicts antenna performance based on the topology of the antenna. Subsequently, the SM replaces EM simulations within the RL training environment. The RL agent interacts with the CNN-based SM to search for the optimal topology. This approach significantly reduces dependence on time-consuming EM simulations. To validate the effectiveness of the optimization method, a center-fed microstrip patch antenna is optimized. Simulation results demonstrate that, compared to other optimization methods, impedance bandwidth is improved, while the number of simulation samples and optimization time are significantly reduced. © ACES.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

USDE-Based Task-Space Bilateral Teleoperator Control Design with Time Delay 16th

USDE-Based Task-Space Bilateral Teleoperator Control Design ...

引用

16th International Conference on Modelling, Identification and Control, ICMIC 2024

作者： Luo, Xin Ma, Jun Wang, Xian Na, Jing Xing, Yashan Faculty of Information Engineering and Automation Kunming University of Science and Technology Kunming650500 China Faculty of Mechanical and Electrical Engineering Kunming University of Science and Technology Kunming650500 China Yunnan Key Laboratory of Intelligent Control and Application Kunming University of Science and Technology Kunming650500 China

ISBN: (纸本)9789819617760

In this paper, we present a control strategy for bilateral teleoperators operated in the task space, which is designed to estimate the uncertain dynamics. To implement this proposed strategy, we develop a straightforward yet robust unknown system dynamics estimator (USDE) by incorporating a low-pass filtering operation. By using the USDE, we can eliminate the impact of unmodeled system dynamics without the need to measure the acceleration signal. Furthermore, the proposed USDE offers the advantages of requiring only one tuning parameter and being easily adjustable compared to conventional estimators. The convergence of the estimation error associated with the unmodeled system dynamics and the stability of the closed-loop control system are both rigorously demonstrated through Lyapunov’s theorem. Ultimately, the superior performance of our method, as evidenced by enhancing motion synchronization and stability, is confirmed through simulation. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

关键词： Time delay

来源：评论

学校读者我要写书评

暂无评论

Soft Sensor Development Based on Large-Scale Pseudo Labeling Optimization 13

Soft Sensor Development Based on Large-Scale Pseudo Labeling...

引用

13th IEEE Data Driven Control and Learning Systems Conference, DDCLS 2024

作者： Liu, Guangkun Jin, Huaiping Wang, Bin Yang, Biao Qian, Bin Kunming University of Science and Technology Faculty of Information Engineering and Automation Department of Automation Kunming650500 China Kunming University of Science and Technology Yunnan Key Laboratory of Artificial Intelligence Kunming650500 China

ISBN: (纸本)9798350361674

Data-driven soft sensor techniques enable online prediction of key quality variables. However, in real industry, the expensive cost of manual labeling leads to the abundance of unlabeled data and the scarcity of labeled data, which becomes a major obstacle to establishing high-precision data-driven soft sensing. To address this problem, a method based on large-scale pseudo-labeling optimization, LSPLO, is proposed. This method, by introducing confidence assessment, fixes the pseudo labels assessed as high-confidence during optimization, while the low-confidence pseudo labels are used as new decision variables after dimensionality reduction to continue optimization. The original decision variables are reinitialized using superior pseudo labels, and stepwise dimensionality reduction and optimization are performed to obtain the optimal pseudo labels. The superiority of the proposed method LSPLO is demonstrated by a real industrial case of the Chlortetracycline (CTC) fermentation process. © 2024 IEEE.

关键词： Dimensionality reduction

来源：评论

学校读者我要写书评

暂无评论

A Deep Learning Soft Sensor Based on Domain-Invariant Features Extraction and Online Local Adaptation 13

A Deep Learning Soft Sensor Based on Domain-Invariant Featur...

引用

13th IEEE Data Driven Control and Learning Systems Conference, DDCLS 2024

作者： Wang, Yelei Jin, Huaiping Wang, Bin Yang, Biao Kunming University of Science and Technology Faculty of Information Engineering and Automation Department of Automation Kunming650500 China Kunming University of Science and Technology Yunnan Key Laboratory of Artificial Intelligence Kunming650500 China

ISBN: (纸本)9798350361674

Data-driven approaches, especially those based on deep learning, have received much attention in quality prediction in the modern process industry. However, in practice, industrial data are often generated in the form of data streams. As operating conditions and process parameters change with the environment, industrial data are dynamic, resulting in degraded model prediction performance. Although the method based on deep transfer learning can automatically extract domain-invariant features in various tasks to adapt to data distribution changes. However, current deep transfer learning methods cannot handle online data streams with dynamic distribution change characteristics. To solve this problem, a deep learning soft sensor method based on domain-invariant features and online local adaptation (DIF-OLA) is proposed. The implementation of the method consists of an offline phase and an online phase. In the offline phase, domain-invariant features are extracted to build a domain-adaption model. In the online phase, a local fine-tuned update strategy is introduced to make the prediction model better adapted to the new data streams. The effectiveness of the method is verified by the dataset of an actual industrial debutanizer column process. The experimental results indicate that compared to traditional adaptation soft sensor methods, the model has improved performance metrics of MAE, RMSE and R2 by 41.14%,40.94%, and 30.94% respectively. © 2024 IEEE.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

End-to-end Relation-Enhanced Learnable Graph Self-attention Network for Knowledge Graphs Embedding 18th

End-to-end Relation-Enhanced Learnable Graph Self-attention ...

引用

18th CCF Conference on Computer Supported Cooperative Work and Social Computing, ChineseCSCW 2023

作者： Jiang, Shengchen Wang, Hongbin Hou, Xiang Faculty of Information Engineering and Automation Kunming University of Science and Technology Kunming650500 China College of Automation Chongqing University Chongqing400044 China

ISBN: (纸本)9789819996360

The knowledge graphs embedding performance of the classic graph convolutional network has been limited due to the large-scale knowledge information. The complex knowledge information requires the model for better learnability rather than linearly weighted qualitative constraints. By studying the structural characteristics of the knowledge graph and investigation the imbalance of knowledge information, the end-to-end relation-enhanced learnable graph self-attention network for knowledge graphs embedding is proposed in this work. A relation-enhanced adjacency matrix is constructed to take into account the incompleteness of the knowledge graph. The convolutional knowledge sub-graph is introduced by using a graph self-attention network to obtain the entity node information’s global encoding and relevance ranking. The training effect of the Convolutional Knowledge Base (convKB) model is improved by changing the construction of negative samples, and a better reliability score in the decoder can be obtained. The experimental results on the data set FB15k-237 and WN18RR show that the proposed method achieves better scores than the compared methods in terms of Hits@10 and MRR. © 2024, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

关键词： Knowledge graph

来源：评论

学校读者我要写书评

暂无评论

Soft Sensor Method based on Quality-related Virtual Sample Generation and Sample-weighted Learning 13

Soft Sensor Method based on Quality-related Virtual Sample G...

引用

13th IEEE Data Driven Control and Learning Systems Conference, DDCLS 2024

作者： Dong, Shuang Jin, Huaiping Wang, Bin Yang, Biao Liu, Haipeng Kunming University of Science and Technology Faculty of Information Engineering and Automation Department of Automation Kunming650500 China Kunming University of Science and Technology Yunnan Key Laboratory of Artificial Intelligence Kunming650500 China

ISBN: (纸本)9798350361674

In process industry, data-driven soft sensor often faces the problem of data shortage in modeling due to factors such as high cost of label samples acquisition and high data repetition rate. The virtual sample generation (VSG) method has been proposed for data augmentation to solve the above problems. Most of the conventional generative models cannot generate virtual samples with labeled data, at the same time, previous data augmentation methods have ignored the quality differences of the generated virtual samples themselves. Thus, this paper proposes a soft sensor method based on quality-related virtual sample generation and sample-weighted learning (QRVSG-SWL). Firstly, this method combines the respective advantages of variational autoencoder (VAE) and generative adversarial network (GAN) to generate labeled virtual samples. Secondly, a prediction model is constructed using virtual samples to calculate the prediction accuracy and distribution differences on real labeled data. Then, sample similarity calculation under supervised latent *** performed. Finally, model learns virtual sample weights. The effectiveness of the proposed method is validated by the industrial chlortetracycline (CTC) fermentation process. © 2024 IEEE.

关键词： Generative adversarial networks

来源：评论

学校读者我要写书评

暂无评论

Improved Deeplabv3+ Method for the Panax Notoginseng Disease Segmentation 11

Improved Deeplabv3+ Method for the Panax Notoginseng Disease...

引用

11th International Conference on information Systems and Computing technology, ISCTech 2023

作者： Lei, Lian Wang, Zilong Wang, Ruoxi Shen, Tao Yang, Qiliang Yang, Ling Kunming University of Science and Technology Faculty of Information Engineering and Automation Kunming China Kunming University of Science and Technology Faculty of Modern Agricultural Engineering Kunming China

ISBN: (纸本)9798350342406

Panax notoginseng diseases seriously affect Panax notoginseng yield and quality. The Panax notoginseng diseases precise identification and segmentation can provide the basis for disease control and treatment. Deep learning technologies gradually displace manual plant disease identification techniques due to their time-consuming, labor-intensive, inefficient, and subjective disadvantages. Consequently, it is essential to explore the potential application of deep learning-based semantic segmentation models in Panax notoginseng disease monitoring. However, due to the complexity of symptoms, ranging from small spots to large-scale diseases, achieving accurate disease segmentation is challenging. To address the above problem, we propose an improved Deeplabv3+ model for Panax notoginseng disease segmentation. Two modules are introduced to improve the performance of the algorithm. Specially, a Dual Attention Module (DAM) is added to improve multiscale feature extraction for achieving accurate segmentation of small disease regions. Moreover, we introduce a Squeeze and Excitation Module (SEM) into the backbone network, which can enhance the ability of feature extraction for large disease regions. Finally, a segmentation dataset of Panax notoginseng diseases is constructed to provide a benchmark for the segmentation of Panax notoginseng diseases. Experimental results show that our method can achieve 92.60% PA and 73.22% mIoU. This study can offer a valuable guide for future monitoring of Panax notoginseng diseases in cultivation trials. © 2023 IEEE.

关键词： Semantic Segmentation

来源：评论

学校读者我要写书评

暂无评论

DDT-Net:Deep Detail Tracking Network for Image Tampering Detection

引用

Computers, Materials & Continua 2025年第5期83卷 3451-3469页

作者： Jim Wong Zhaoxiang Zang Hubei Key Laboratory of Intelligent Vision Based Monitoring for Hydroelectric Engineering China Three Gorges UniversityYichang443002China College of Computer and Information Technology ChinaThree Gorges UniversityYichang443002China Faculty of Information Engineering and Automation Kunming University of Science and TechnologyKunming650504China

In the field of image forensics,image tampering detection is a critical and challenging *** methods based on manually designed feature extraction typically focus on a specific type of tampering operation,which limits their effectiveness in complex scenarios involving multiple forms of *** deep learningbasedmethods offer the advantage of automatic feature learning,current approaches still require further improvements in terms of detection accuracy and computational *** address these challenges,this study applies the UNet 3+model to image tampering detection and proposes a hybrid framework,referred to as DDT-Net(Deep Detail Tracking Network),which integrates deep learning with traditional detection *** contrast to traditional additive methods,this approach innovatively applies amultiplicative fusion technique during downsampling,effectively combining the deep learning feature maps at each layer with those generated by the Bayar noise *** design enables noise residual features to guide the learning of semantic features more precisely and efficiently,thus facilitating comprehensive feature-level ***,by leveraging the complementary strengths of deep networks in capturing large-scale semantic manipulations and traditional algorithms’proficiency in detecting fine-grained local traces,the method significantly enhances the accuracy and robustness of tampered region *** with other approaches,the proposed method achieves an F1 score improvement exceeding 30% on the DEFACTO and DIS25k *** addition,it has been extensively validated on other datasets,including CASIA and *** results demonstrate that this method achieves outstanding performance across various types of image tampering detection tasks.

关键词： Image forensics image tampering detection image manipulation detection noise flow Bayar

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

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

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

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

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：