检索结果-内蒙古大学图书馆

2022 Global Conference on Robotics, Artificial Intelligence and information Technology, GCRAIT 2022

作者： Hua, Wendi Xie, Ping School of Computer Qinghai Normal University Key Laboratory of Internet of Things of Qinghai Province Qinghai China State Key Laboratory of Tibetan Intelligent Information Processing and Application Institute of Plateau Science and Sustainability Xining China

ISBN: (数字)9781665481922

ISBN: (纸本)9781665481922

Cuckoo filter is a new type of approximate membership query data structure that stores the fingerprints of set elements. Because of its higher performance than bloom filter, cuckoo filter has been developed in recent years and gradually replaced the bloom filter in many fields. However, the endless loop problem in cuckoo hashing makes element insertion inefficient and unstable, which hinders cuckoo filter development. To solve this problem, we propose a new variant of cuckoo filter called smart cuckoo filter. Smart cuckoo filter uses a variable bucket structure to save the element insertion information at each position. It keeps buckets aligned to memory, thereby reducing the space waste and inefficiency of data access. Smart cuckoo filter uses the disjoint-set to determine whether the element will fall into the endless loop before the element is inserted. Also, the insertion is friendly to the concurrency of the entire filter. The experiments demonstrate that the element insertion speed of smart cuckoo filter in the endless loops is better than that of the traditional cuckoo filter, which can reach 2.2x in some cases. © 2022 IEEE.

关键词： Data structures

来源：评论

学校读者我要写书评

暂无评论

TD-H:A Method for High-Quality Computer-Generated Holograms 10

TD-H:A Method for High-Quality Computer-Generated Holograms

引用

10th International Conference on Systems and Informatics, ICSAI 2024

作者： Wang, Zhiguo Dong, Xiuyu Zuo, Jing Geng, Shengling Qinghai Normal University The State Key Laboratory of Tibetan Intelligent Information Processing and Application Xining China School of Computer Science Qinghai Normal University Xining China Beijing Normal University Qinghai Normal University Academy of Plateau Science and Sustainability People's Government Of Qinghai Province Xining China

ISBN: (纸本)9798331510138

Traditional computer-generated holography (CGH) algorithms rely on complex diffraction calculations, resulting in high computational demands and low efficiency. To generate high-quality holograms more rapidly, a deep learning approach simplifies the hologram generation process. The innovative method, Generate Holograms through Two Deep Neural Networks (TD-H), efficiently produces high-quality holograms by integrating two deep neural networks: the Attention Dense Prediction Transformer (ADPT) and the U-Net Model with Attention Mechanism (UAN). ADPT introduces an attention mechanism to predict depth information from the RGB values of images, while UAN combines the U-Net architecture with Multi-Head Attention (MHA) to efficiently generate high-quality holograms using both RGB values and the predicted depth map. On the DML/CVL RGB-D dataset, the average Peak Signal-to-Noise Ratio (PSNR) achieved by the ADPT and UAN models is 39.01 dB and 31.54 dB, respectively, with the ADPT model outperforming the second-ranking Dense Prediction Transformer (DPT) model by 3% in PSNR. The TD-H method not only simplifies the hologram generation process but also accurately converts and produces high-quality holograms, effectively addressing the issues of low efficiency and poor quality in hologram generation while representing natural reconstructed images in the depth direction. © 2024 IEEE.

关键词： Computer generated holography

来源：评论

学校读者我要写书评

暂无评论

Classification of Underwater Fish Image Based on Supervised Contrastive Learning

Classification of Underwater Fish Image Based on Supervised ...

引用

intelligent Robotics and Systems (ISoIRS), International Symposium on

作者： Qiankun Yan Xiujuan Du College of Computer Qinghai Normal University Qinghai Provincial Key Laboratory of IoT Xining China The State Key Laboratory of Tibetan Intelligent Information Processing and Application Xining China

ISBN: (数字)9798350364798

ISBN: (纸本)9798350364804

Classification of underwater fish images is of great scientific research significance for assessing the ocean's health. However, compared with land-based optical images, underwater optical images are characterized by color distortion, low contrast, and high noise intensity, making it challenging to perform effective feature extraction and train a highly accurate recognition model for underwater images. This paper proposes a supervised contrastive learning-based classification method for underwater fish images. First, the residual idea is integrated to improve the representation of the last four stages of the residual network to obtain the stage residual network, which more fully utilizes the shallow features and reduces the risk of network overfitting at the same time; second, the channel attention mechanism is incorporated at the end of the last four stages of the backbone network, which prompts the model to pay more attention to the feature channels with high contribution; finally, the contrastive learning loss function is improved by the introduction of labels, and realizes the supervised contrastive learning. Dividing the backbone network and the classifier into two stages of training effectively closes the intra-class distance and distances the inter-class distance and improves the model's performance. The experimental results show that the accuracy, precision, recall, and F1 value of this paper's model on the public benchmark dataset Fish4Knowledge are 99.28%, 98.17%, 94.15%, and 95.71%, and it can realize the classification of underwater fish images under extreme noise conditions more efficiently.

关键词： Training Accuracy Oceans Contrastive learning Optical distortion Fish Optical imaging

来源：评论

学校读者我要写书评

暂无评论

Error Correction of tibetan Verbs Based on Deep Learning 2

Error Correction of Tibetan Verbs Based on Deep Learning

引用

2nd International Conference on Artificial Intelligence and Computer Engineering, ICAICE 2021

作者： Hua, Guo-Cai-Rang Jia, Secha Ban, Ma-Bao Cai, Rang-Jia Qinghai Key Laboratory of Tibetan Information Processing and Machine Translation 810008 China Qinghai Tibetan Information Processing Engineering Technology Research Center 810008 China State Key Laboratory of Tibetan Intelligent Information Processing and Application 810008 China

ISBN: (纸本)9781665421867

Verbs are the core of the semantic structure represented by tibetan sentences, and automatic error correction of tibetan verbs is one of the important research topics in tibetan language processing. In this paper, we propose a Bi-LSTM neural network model for tibetan verb error correction by analyzing the usage rules of verbs in tibetan, summarizing the grammatical, semantic and spelling features of verbs, and proposes a Bi-LSTM neural network model for tibetan verb error correction based on these features, which not only extracts various features of verbs, but also capture the contextual information of verbs through Bi-LSTM neural network. The method proposed in this work solves the drawbacks of the traditional methods of low generalization and inability to obtain long-distance contextual implicit information. The experimental results show that the accuracy, recall and F1 values of the proposed method on the test set reach 97.3%,95.7% and 96.9%, respectively, indicating the effectiveness of the proposed method on the task of automatic error correction for tibetan verbs. © 2021 IEEE.

关键词： Error correction

来源：评论

学校读者我要写书评

暂无评论

Energy Consumption Optimization of an IoT Monitoring Center Based on a Max-Min Ant Colony Algorithm

引用

Wireless Communications and Mobile Computing 2023年第1期2023卷

作者： Tong, Yinghua School of Computer Qinghai Normal University Xining810008 China The State Key Laboratory of Tibetan Intelligent Information Processing and Application Xining810008 China

As a result of the green IoT, application layer users are increasingly seeking more services, and it is particularly important to reduce the energy consumption of an IoT monitoring center while ensuring service reliability. In this paper, the max-min ant colony algorithm is proposed to effectively solve the problems of user service reliability and energy consumption optimization of an IoT monitoring center. First, the service reliability of the application layer users is modeled based on a Markov chain, and the quantitative reliability performance indicators are given. Second, an optimization model between the user service reliability and the energy consumption of an IoT monitoring center is established. Finally, the optimization problem is solved based on the improved max-min ant system. Simulation experimental results show that compared with the improved FFD algorithm, the average power consumption of the max-min ant colony algorithm can be reduced by up to 7.14%. Copyright © 2023 Yinghua Tong.

关键词： Energy utilization

来源：评论

学校读者我要写书评

暂无评论

Multi Strategy Improvement of Newton-Raphson-Based Optimizer and Engineering application

Multi Strategy Improvement of Newton-Raphson-Based Optimizer...

引用

International Conference on Machine Learning, Big Data and Business Intelligence (MLBDBI)

作者： Shaogang Cheng Jun Yin Tingting Liu College of Computing Qinghai Normal University Xining China The State Key Laboratory of Tibetan Intelligent Information Processing and Application Qinghai Normal University Xining China

ISBN: (数字)9798331541798

ISBN: (纸本)9798331505615

A multi strategy improved Newton- Raphson-Based Optimizer (NRBO) is proposed to address the problem of slow convergence speed in the later stage of algorithm iteration and difficulty in finding the global optimal solution when facing complex functions. Firstly, using a set of best points instead of randomly generated initial populations to evenly distribute individuals in the optimization space improves the efficiency of the algorithms global and local search. Secondly, based on the idea of differential evolution, the initial population is mutated and crossed to accelerate the iteration speed. Then, in the optimal update stage, the Levi flight strategy is introduced to enhance the algorithms ability to jump out of local optima. The effectiveness of the proposed algorithm was tested on the CEC2005 test set, and compared with other algorithms on two test sets. The results showed that the convergence speed and accuracy of the algorithm were significantly improved. Finally, the proposed algorithm was applied to two practical engineering optimization problems, and simulation results showed that the algorithm proposed in this paper has better practicality and robustness compared to other algorithms.

关键词： Machine learning algorithms Accuracy Simulation Machine learning Robustness Business intelligence Optimization Faces Convergence Testing

来源：评论

学校读者我要写书评

暂无评论

YOLOX-TE: Remote Sensing Image Object Detection Based on Improved YOLOX-Tiny 16

YOLOX-TE: Remote Sensing Image Object Detection Based on Imp...

引用

16th International Conference on Advanced Computer Theory and Engineering, ICACTE 2023

作者： Xie, Yunchuan Geng, Shengling Zhang, Dan Wang, Fubo Wang, Yuxiang Yan, Yuhang School of Computer Science Qinghai Normal University Qinghai Xining810016 China The State Key Laboratory of Tibetan Intelligent Information Processing and Application Qinghai Xining810008 China Academy of Plateau Science and Sustainability Qinghai Normal University Qinghai Xining810008 China

ISBN: (纸本)9798350317534

Lightweight object detection models have been widely applied in remote sensing image object detection and have achieved good results. However, there are still some unresolved issues in lightweight remote sensing image object detection, such as dense target distribution, small size, and the inability of the network to effectively extract deep feature information of the target, which can easily lead to issues such as false detections, missed detections, and inaccurate boundary box positioning. Therefore, we propose YOLOX-TE. It can effectively extract deep features of small target objects and enhance the modeling ability of target bounding boxes at different scales. Firstly, it optimizes the feature pyramid network FPN through inverse Focus operation and transposed convolution. Secondly, a multi-scale fusion module based on ECA attention mechanism is added to the head network. Finally, an improved SENet attention mechanism is added to the bounding box regression branch. According to the RSOD dataset validation, compared to the YOLOX-Tiny model, the YOLOX-TE model only increased the parameter count by 0.363M, but the accuracy of AP, AP75, APS, and APL increased by 7.1%,6.7%, 6.8%, and 7.3%, respectively. Compared with the most advanced lightweight YOLOV7-Tiny model, the accuracy of AP and AP75 is 0.4% and 2.1% higher, respectively, and YOLOX-TE has a smaller volume and lighter weight. © 2023 IEEE.

关键词： Object detection

来源：评论

学校读者我要写书评

暂无评论

Bistreaming Attention Network for Reproducible Speech Emotion Recognition

SSRN

引用

SSRN 2024年

作者： Zhang, Huiyun Huang, Heming School of Computer Science Qinghai Normal University Xining810008 China The State Key Laboratory of Tibetan Intelligent Information Processing and Application Xining810008 China

Speech emotion recognition (SER) plays a pivotal role in enhancing the fluidity of Human-computer interaction (HCI) applications. Increasingly, researchers are turning to deep learning models to tackle SER tasks. However, this transition poses challenges, particularly in terms of the effectiveness and reproducibility of these models. To accurately assess reproducibility, we establish a comprehensive definition and employ specific measures. Additionally, we introduce a novel SER model, Bistreaming attention (BSA), with distinct layers: an input layer, an isomorphic multi-head attention layer, a fusion layer, and an output layer. Our findings reveal that the proposed BSA model exhibits significant advantages over its counterparts in tackling SER tasks across diverse datasets, including EmoDB, CASIA, SAVEE, BodEmoDB, IEMOCAP, and ESD. Specifically, the BSA model showcases superior accuracy and consistency in emotion label prediction over multiple experiments, signifying heightened robustness, generalization, and reproducibility compared to peer models. The BSA model addresses deep learning challenges in SER by innovatively capturing speech emotion patterns, leveraging BSA model to focuses on relevant acoustic features. With its superior performance and reproducibility, the BSA model emerges as a promising solution for advancing SER research, paving the way for enhanced HCI applications. © 2024, The Authors. All rights reserved.

关键词： Human computer interaction

来源：评论

学校读者我要写书评

暂无评论

A tibetan Ancient Uchen Text Line Dataset for OCR

A Tibetan Ancient Uchen Text Line Dataset for OCR

引用

Image processing, Computer Vision and Machine Learning (ICICML), International Conference on

作者： Zhuome Gongqu Peng Luo Dongzhou Jiayang Jia Cairang Jiacuo Cizhen Dongzhu Renqing The College of Computer Qinghai Normal University Xining China The State Key Laboratory of Tibetan Intelligent Information Processing and Application Xining China Tibetan Information Processing and Machine Translation Key Laboratory of Qinghai Province Xining China The College of Information Science and Technology Tibet University Lhasa China

ISBN: (数字)9798350355413

ISBN: (纸本)9798350355420

Characters written by hand that represent distinct writing styles are referred to as handwritten text; these are especially common in ancient texts. In the field of digitizing ancient literature, tibetan ancient document image recognition is a focus of both domestic and international research. However, technological advancement in this field has been hampered by the absence of a specific recognition dataset for image recognition systems of tibetan ancient documents. This paper addresses this by proposing a dataset for Uchen style text line recognition in tibetan ancient documents. This dataset includes basic synthetic images of multi-style Uchen text lines as well as real-world images of multi-format tibetan Uchen text lines that have been manually annotated. Cross-Validation and several rounds of annotation by various people guarantee the quality of the data. Additionally, three popular text line recognition models are used to assess the dataset and validate its usability.

关键词： Hands Computer vision Image recognition Text recognition Annotations Computational modeling Optical character recognition Machine learning Writing Usability

来源：评论

学校读者我要写书评

暂无评论

Synchronization analyze of k-uniform hyper-networks (vol 14, 6125 , 2024)

引用

SCIENTIFIC REPORTS 2024年第1期14卷 1-15页

作者： Du, Juan Ma, Xiujuan Ma, Fuxiang Yu, Wenqian School of Computer Science Qinghai Normal University Xining 810000 Qinghai China The State Key Laboratory of Tibetan Intelligent Information Processing and Application Qinghai Normal University Xining 810016 Qinghai China

Hyper-networks tend to perform better in representing multivariate relationships among nodes. Yet, due to the complexity of the hyper-network structure, research in synchronization dynamics is rarely involved. In this paper, a Kuramoto model more suitable fork-uniform hyper-networks is proposed. And the generalized Laplacian matrix expression of thek-uniform hyper-network is present. We use the eigenvalue ratio of the generalized Laplacian matrix to quantify synchronization. And we studied the effects of some important structure parameters on the synchronization of three types ofk-uniform hyper-networks. And obtained different relationships between synchronization and these parameters. The results show the synchronization of thek-uniform hyper-networks is related to both structure and parameters. And as the size of the nodes increases, the synchronization ability gradually increases for ER random hyper-network, while that gradually decreases for NW small-world hyper-network and BA scale-free hyper-network. As the uniformity increases, the synchronization ability of all three types of uniform hyper-networks increases. In addition, when the structure and node size are fixed, the synchronization ability increases with the increase of the hyper-clustering coefficient in BA scale-free hyper-network and ER random hyper-network, while it decreases with the increase of the hyper-clustering coefficient in NW small-world hyper-network.

关键词：

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：