检索结果-内蒙古大学图书馆

您好，读者！请登录

内蒙古大学图书馆

首页
概况
党建
资源
服务
科研支持
- 论文收录引用证明
- 科技查新
知识产权
档案馆
帮助

咨询与建议

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

您的常用邮箱：*

您的手机号码：*

问题描述：

当前已输入0个字，您还可以输入200个字

全部搜索
期刊论文
图书
学位论文
标准
纸本馆藏
外文资源发现
数据库导航
超星发现

高级检索

时间限定

出版年份：

文献类型

图书期刊文献学位论文多媒体

馆藏选择

电子馆藏纸本馆藏

核心期刊

全部期刊 SCI 收录期刊 SSCI 收录期刊 EI 收录期刊 CSCD 收录期刊 CSSCI 收录期刊

语言

中文英文

文献类型

期刊文献图书学位论文标准纸本馆藏

帮助

文字说明：

T=题名（书名、题名），A=作者（责任者），K=主题词，P=出版物名称，PU=出版社名称，O=机构（作者单位、学位授予单位、专利申请人），L=中图分类号，C=学科分类号，U=全部字段，Y=年（出版发行年、学位年度、标准发布年）

检索规则说明：

AND代表“并且”；OR代表“或者”；NOT代表“不包含”；(注意必须大写,运算符两边需空一格)

检索范例：

范例一：(K=图书馆学 OR K=情报学) AND A=范并思 AND Y=1982-2016
范例二：P=计算机应用与软件 AND (U=C++ OR U=Basic) NOT K=Visual AND Y=2011-2016

分类表

所选分类

>> <<

限定检索结果

文献类型

2,645 篇 会议
932 篇 期刊文献
36 册 图书

馆藏范围

3,613 篇 电子文献
0 种 纸本馆藏

日期分布

学科分类号

2,007 篇 工学
- 1,536 篇 计算机科学与技术...
- 1,260 篇 软件工程
- 558 篇 信息与通信工程
- 399 篇 控制科学与工程
- 262 篇 电气工程
- 210 篇 生物工程
- 206 篇 电子科学与技术（可...
- 196 篇 网络空间安全
- 188 篇 生物医学工程（可授...
- 156 篇 光学工程
- 149 篇 动力工程及工程热...
- 145 篇 安全科学与工程
- 125 篇 化学工程与技术
- 123 篇 机械工程
- 111 篇 交通运输工程
- 94 篇 仪器科学与技术
1,123 篇 理学
- 570 篇 数学
- 325 篇 物理学
- 254 篇 生物学
- 199 篇 统计学（可授理学、...
- 141 篇 化学
- 117 篇 系统科学
615 篇 管理学
- 374 篇 管理科学与工程(可...
- 268 篇 图书情报与档案管...
- 196 篇 工商管理
270 篇 医学
- 196 篇 临床医学
- 187 篇 基础医学(可授医学...
- 167 篇 公共卫生与预防医...
- 103 篇 药学(可授医学、理...
129 篇 法学
- 105 篇 社会学
69 篇 农学
63 篇 经济学
25 篇 教育学
15 篇 文学
8 篇 军事学
5 篇 艺术学

主题

229 篇 accuracy
151 篇 deep learning
131 篇 machine learning
129 篇 real-time system...
114 篇 computational mo...
113 篇 convolutional ne...
111 篇 internet of thin...
110 篇 feature extracti...
96 篇 predictive model...
86 篇 support vector m...
77 篇 training
71 篇 optimization
70 篇 technological in...
67 篇 cloud computing
67 篇 data models
66 篇 artificial intel...
63 篇 security
57 篇 machine learning...
53 篇 wireless sensor ...
49 篇 scalability

机构

115 篇 lovely professio...
106 篇 department of co...
92 篇 department of co...
64 篇 gla university d...
57 篇 department of co...
53 篇 lloyd law colleg...
47 篇 department of co...
42 篇 lloyd institute ...
42 篇 the islamic univ...
42 篇 school of comput...
40 篇 beijing engineer...
39 篇 school of comput...
39 篇 chitkara univers...
36 篇 graphic era deem...
35 篇 gla university d...
35 篇 institute of aer...
34 篇 department of co...
32 篇 graphic era deem...
30 篇 saveetha school ...
30 篇 lloyd institute ...

作者

47 篇 neeraj varshney
41 篇 varshney neeraj
37 篇 rakesh kumar
33 篇 singh ashutosh k...
33 篇 kumar rakesh
31 篇 huang heyan
28 篇 saloni bansal
27 篇 singh kamred udh...
26 篇 asha v.
25 篇 bibhuti bhusan d...
25 篇 singh teekam
25 篇 sudhansu shekhar...
23 篇 saxena deepika
22 篇 kumar ankit
22 篇 patra sudhansu s...
20 篇 n. naga saranya
20 篇 kumar gaurav
20 篇 shrivastava anur...
20 篇 anurag shrivasta...
19 篇 bansal saloni

语言

2,550 篇 英文
1,040 篇 其他
33 篇 中文
1 篇 德文
1 篇 法文

检索条件"机构=Institute for Computer Applications in Science and engineering"

共 3613 条记录，以下是301-310 订阅

全选清除本页清除全部题录导出标记到"检索档案"

详细简洁

排序：

Crop Classification on Hyperspectral Images Using Residual Network and Multi-Head Self-Attention Feature Extraction 2

Crop Classification on Hyperspectral Images Using Residual N...

引用

2nd IEEE International Conference on Data science and Network Security, ICDSNS 2024

作者： Saranya, N. Naga Vivekanandhan, V. Al-Farouni, Mohammed Shreyas, A.V. Kumar, Gotte Ranjith Saveetha College of Liberal Arts & Science Simats Department of Computer Applications Chennai India Malla Reddy College of Engineering Department of Computer Science and Engineering Hyderabad Secunderabad India College of Technical Engineering The Islamic University Department of Computers Techniques Engineering Najaf Iraq Nitte Meenakshi Institute of Technology Department of Civil Engineering Bengaluru India School of Computer Science & Artificial Intelligence Sr University Warangal India

ISBN: (纸本)9798350373110

Nowadays, hyperspectral imaging has empowered classification and prediction in many fields such as agriculture, land usage, tourism, etc. due to advancements in deep learning. This research focuses on obtaining efficient crop classification using Residual Network, Multi-Head Self- Attention, Convolutional Neural Network (ResNet-MHSA- CNN) approach. The Adaptive Wiener Filtering (AWF) is applied in preprocessing to eliminate noise, handles blurred parts of images effectively and decreases error rate in inverse filtering. The Residual Network (ResNet) method is employed by implement residual learning to eliminate degradation issues of deep neural networks. The high-resolution depth features of hyperspectral images are extracted by residual learning ResNet network. The Multi-Head Self-Attention (MHSA) utilized to capture long-range dependencies and global context information and helps the model in learning diverse patterns. Finally, the extracted features are given to Convolutional Neural Network (CNN) with complex pattern learning ability that excels in image classification with multiple layers of filters. The proposed ResNet-MHSA-CNN approach is evaluated with Indian Pines dataset and achieved 99.74%, 99% accuracy and kappa value respectively. The proposed method outperformed state-of-art methods such as Cross-Mixing Residual Network of Convolutional Neural Network (CMR-CNN), Deep Stacked Denoising Autoencoder (DSDA). © 2024 IEEE.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Contrastive Loss Function Siamese Network for Change Detection on Remote Sensing Images 2

Contrastive Loss Function Siamese Network for Change Detecti...

引用

2nd IEEE International Conference on Data science and Network Security, ICDSNS 2024

作者： Saranya, N. Naga Alkhayyat, Ahmad Shreyas, A.V. Raviprakash, M.L. Jayachandran, T. Saveetha College of Liberal Arts & Science Simats Department of Computer Applications Chennai India College of Technical Engineering The Islamic University Department of Computers Techniques Engineering Najaf Iraq Nitte Meenakshi Institute of Technology Department of Civil Engineering Bengaluru India Kalpataru Institute of Technology Department of Computer Science and Engineering Tiptur India Nandha Engineering College Department of Ece Erode India

ISBN: (纸本)9798350373110

To observe the changes in the earth surface is increasing due to development of navigation software applications and also mainly prediction purposes for agriculture purpose and environmental hazards. To overcome the problem one of the most crucial technological choices for effectively identifying changes in observations in remote sensing photos is Change Detection (CD). applications for CD are numerous and include mapping disasters, environmental monitoring, urban planning and land use study. On the other hand, the CD applications face significant hurdles due to the often-occurring class imbalance issue. In order to tackle this problem, a new type of Contrastive loss function Siamese network (CLFS-Net) that integrates the links between scenes and foregrounds to yield global-aware features for effective change detection. The CLFS-Net improve symbiotic connection learning between scene and foreground while simultaneously learning contextual relationships. The proposed CLFS-Network gained in terms of accuracy as 90%, precision as 95% and recall as 91% and f1 score as 95% respectively. © 2024 IEEE.

关键词： Environmental monitoring

来源：评论

学校读者我要写书评

暂无评论

Recognition of Human Activity using Signal Processing & Deep Neural Networks 8

Recognition of Human Activity using Signal Processing & Deep...

引用

8th IEEE International Conference for Convergence in Technology, I2CT 2023

作者： Moharana, Bhimasen Biswal, Bhramara Bar Samant, Tapaswini Behera, Trupti Mayee Mishra, Subhashree Banerjee, Shobhan Lovely Professional University Department of Computer Science & Engineering Punjab India College of Engineering P.G. Department of Computer Science & Applications Bhubaneswar India Kalinga Institute of Industrial Technology School of Electronics Engineering Bhubaneswar India Birla Institute of Technology and Science - Pilani Department of Engineering Technology Rajasthan India

ISBN: (纸本)9798350334012

The task of tracking the activities being performed by humans is of immense importance as it finds application in healthcare, marketing, surveillance, etc. To keep track of this, they need to be recognized and classified into various activities. The data can be collected through smart devices such as smartwatches or smartphones which internally use various sensors to measure the physical movements of the person using it. We have used Human Activity Recognition Dataset from the UCI ML Repository, which consists of the data being collected from the smartphones of 30 volunteers. After choosing the appropriate features, we trained an Artificial Neural Network and a Long Short Term Memory Neural Network to classify the activities into various categories and performed a comparative analysis of these neural networks among themselves & also with respect to the previously leveraged techniques & algorithms on the dataset. © 2023 IEEE.

关键词： Long short-term memory

来源：评论

学校读者我要写书评

暂无评论

Generative Adversarial Networks (GANs) in the Wild: applications, Challenges, and Future Prospects 2

Generative Adversarial Networks (GANs) in the Wild: Applicat...

引用

2nd International Conference on Disruptive Technologies, ICDT 2024

作者： Kumar, Rakesh Chaudhary, Pooja Gupta, Shilpi Singhal, Manoj Gla University Department of Computer Engineering & Applications UP Mathura India UP Ghaziabad India Meerut Institute of Technology Department of Computer Science & Engineering UP Meerut India Gl Bajaj Institute of Technology & Management Department of Information Technology UP Greater Noida India

ISBN: (纸本)9798350371055

Generative adverse Networks (GANs) have become increasingly famous for a variety of tasks in picture and audio processing, especially unsupervised getting to know. GANs encompass components, a 'generator' and a 'discriminator,' that are skilled adversarial. The generator is tasked with producing a pattern of statistics, and the discriminator is tasked with figuring out whether the information is 'actual' or 'fake.' The generator 'learns' through trial and mistakes as it attempts to create samples that appear practical to the discriminator. Through the antagonistic schooling manner, the generator can create more accurate 'fake' records while the discriminator can emerge as better at distinguishing between 'actual' and 'faux' records. GANs have been applied to various duties, which include picture era, textual content to photosynthesis, photo-to-image translation, and automatic music composition. Recent tries have been made to apply GANs to extra complicated obligations in the wild, including self-reliant use and drug discovery. GANs have inspired an ever-developing quantity of industries because of their potential to generate sensible facts with minimal human input. © 2024 IEEE.

关键词： antagonistic complicated discriminator photosynthesis potential

来源：评论

学校读者我要写书评

暂无评论

An Optimized DDoS Attack Detection Using Deep Convolutional Generative Adversarial Networks 5

An Optimized DDoS Attack Detection Using Deep Convolutional ...

引用

5th International Conference on Inventive Research in Computing applications, ICIRCA 2023

作者： Akana, Chandra Mouli Venkata Srinivas Kumar, Arvind Tiwari, Mohit Yunus, Aslanov Zabit Vijayakumar, E. Singh, Mukesh Bonam Venkata Chalamayya Engineering College Odalarevu India School of Computer Applications Manav Rachna International Institute of Research and Studies India Bharati Vidyapeeth's College of Engineering Department of Computer Science and Engineering Delhi India Azerbaijan State University of Economics - Unec Baku Azerbaijan KIT-Kalaignar Karunanidhi Institute of Technology Department of Computer Applications Tamilnadu Coimbatore India Graphic Era University Department of Computer Science and Engineering Dehradun India

ISBN: (纸本)9798350321425

Attacks known as Distributed Denial-of-Service (DDoS) are rising as a result of recent, dramatic increase in demand for Internet access. When the amount and characteristics of network traffic, which may include harmful DDoS contents, expand dramatically, traditional basic algorithms using machine learning for classifying DDoS attacks frequently fail because it cannot automatically extract high-value characteristics. In order to effectively identify and classify DDoS attacks, a hybrid technique called DFNN-SAE-DCGAN that combines three deep learning-based models is suggested. The Deep Feed-Forward Neural Network (DFNN) and Stacked Auto encoder offers an efficient method for extracting features that identifies the most pertinent feature sets without the assistance of a person. To avoid the operational overhead and presumption associated with processing massive set of features with distortion and redundant characteristic values, the Deep Convolutional Generative Adversarial Networks (DCGAN) component of the proposed model classifies the attacks into various DDoS attack types using the restricted and minimized characteristic sets generated by the DFNN-SAE as inputs. The experimental results show a very high and resilient accuracy rate and an F1-score of 98.5%, which is higher than the performance of many similar approaches. These results were acquired by thorough and extensive trials on various performance aspects on the CICDDoS2019 dataset. This demonstrates that the suggested methodology may be utilized to defend against the increasing amount of DDoS attacks. © 2023 IEEE.

关键词： Denial-of-service attack

来源：评论

学校读者我要写书评

暂无评论

Performance Analysis of Classifiers in Predicting Car Insurance Claim 8

Performance Analysis of Classifiers in Predicting Car Insura...

引用

8th International Conference on Inventive Systems and Control, ICISC 202

作者： Bhoji, Spoorthi Pandey, Trilok Nath Dash, Bibhuti Bhusan Satpathy, Rabinarayan De, Utpal Chandra Patra, Sudhansu Shekhar Vellore Institute of Technology School of Computer Science and Engineering Tamil Nadu Chennai India Kiit Deemed to Be University School of Computer Applications Bhubaneswar India Sri Sri University Faculty of Emerging Technologies Cuttack India

ISBN: (纸本)9798350386578

Insurance claims pose a significant and costly challenge for insurance companies. Managing insurance claims presents a fundamental issue for insurance firms. Insurers continually face the challenge of mitigating the escalation of claim losses due to fraudulent activities and the increasing volume of claim data within their operations. Consequently, accurately assessing the status of insured claims during the review process becomes increasingly complex. In order to make the car insurance claim prediction more accurate and feasible a comparison study on the performance of various machine learning and deep learning algorithms is presented in this paper. In this research study, the primary research contribution is that various machine learning and deep learning algorithms like logistic regression, decision trees, random forest, catboost, multilayer perceptron etc. have been analyzed and performance has been evaluated based on the accuracy, precision, recall, F1-score, ROC AUC and confusion matrices. This study shows that some algorithms outperform the others and the Random Forest (RF) algorithm has resulted with the highest accuracy of 98.0%. © 2024 IEEE.

关键词： Logistic regression

来源：评论

学校读者我要写书评

暂无评论

Robustness Challenges in Deep Learning: Strategies for Enhancing Model Resilience 2

Robustness Challenges in Deep Learning: Strategies for Enhan...

引用

2nd International Conference on Disruptive Technologies, ICDT 2024

作者： Joshi, Prakash Shaikh, M.Z. Varshney, Neeraj Dwivedy, Bhoopendra Raj Kumar Goel Institute of Technology Department of Computer Science & Engineering UP Ghaziabad India Shree Bhagubhai Mafatlal Polytechnic College Department of Electronics & Telecommunication Maharastra Mumbai India Gla University Department of Computer Engineering and Applications UP Mathura India Galgotias University Department of School of Computer Science & Engineering UP Greater Noida India

ISBN: (纸本)9798350371055

Deep knowledge is the main way the improvement of gadgets, gaining knowledge of generation for more than a few applications from facial popularity to self-reliant vehicles. Deep learning fashions have carried out outstanding successes, even outperforming people in complex obligations. However, robustness demanding situations of deep getting-to-know systems remain an essential difficulty. Deep studying fashions are at risk of malicious inputs that could result in misclassifications or denial-of-provider attacks. Robustness towards entering noise, in addition to robust extraction of beneficial capabilities from noisy datasets, are just a number of the main challenges for deep mastering systems. Current research has discovered a plethora of attack vectors that can be deployed against deep getting-to-know fashions along with 'adversarial' assaults, which contain manipulating the enter records in a particular manner to purpose models to misclassify items. Moreover, dataset bias and absence of right validation can cause structures that are not sturdy towards facts from a domain extraordinary from the facts used for education. A good way to enhance the robustness of deep learning models, researchers are exploring solutions that include countermeasures, specialized architectures, and higher education and checking out approaches. While progress has been made, the robustness challenges of deep learning remain an important study vicinity that should be addressed so one can fulfill the promise of this generation. © 2024 IEEE.

关键词： Architectures Discovered Extraordinary Manipulating Researchers

来源：评论

学校读者我要写书评

暂无评论

Transfer Learning with ResNet50 for Robust Traffic Sign Recognition: Addressing Challenges in Lighting, Occlusion, and Class Imbalance 2

Transfer Learning with ResNet50 for Robust Traffic Sign Reco...

引用

2nd International Conference on IT Innovations and Knowledge Discovery, ITIKD 2024

作者： Nair, Rekha R Kishore, S. Santhosh, Nandana Babu, Tina Ramamoorthy, H. Alliance School of Advanced Computing Alliance University Department of Computer Science and Engineering Bengaluru India Department of Computer Applications Nitte Institute of Professional Education Mangalore India

ISBN: (纸本)9798350355468

The efficacy of deep learning models, specifically Convolutional Neural Networks (CNNs), in accurately recog- nizing and clas- sifying a diverse range of traffic signs. The research utilizes a meticulously curated dataset obtained from Kaggle, comprising annotated images labeled with corresponding ClassIds representing various traffic sign categories. The moti- vation behind this work stems from the critical need for robust and reliable traffic sign recognition systems, which play a vital role in enhancing safety, efficiency, and enabling autonomy in transportation systems. Through systematic experimentation and rigorous validation, the study assesses the performance of CNN models across various evaluation metrics, shedding light on the effectiveness of these deep learning architectures in accurately discerning nuanced visual cues present in traffic sign imagery. Furthermore, the research explores the challenges posed by factors such as variations in lighting conditions, occlusions, and class imbalances, proposing potential mitigation strategies. The proposed methodology utlized transfer learning model ResNet50 which achieved accuracy of 92. © 2025 IEEE.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Instantaneous Interpretation into Sign Language for the Hearing Impaired 1

引用

3rd International Conference on Universal Threats in Expert applications and Solutions, UNI-TEAS 2024

作者： Singh, Kamred Udham Kumar, Ankit Kumar, Gaurav Singh, Teekam Kothari, Paras Sheikh, Anisha School of Computing Graphic Era Hill University Dehradun India Department of Information Technology Guru Ghasidas Vishwavidyalaya Bilaspur India Department of Computer Engineering and Applications GLA University UP Mathura India Department of Computer Science and Engineering Graphic Era Deemed to Be University Dehradun India Department of Computer Applications Geetanjali Institute of Technical Studies Rajasthan Udaipur India

ISBN: (数字)9789819738106

ISBN: (纸本)9789819738090

Because they are unable to convey their thoughts to others who can hear, deaf persons may have a difficult time controlling their emotions and may find that their routines are often interrupted. This is the primary impetus behind the progression of things. These individuals would be helped by this system, and it would make it possible for them to convey their ideas in a manner that is on par with that of persons who do not have any kind of physical impairment. Recent advances in artificial intelligence have made it possible to design a system that is capable of resolving this problem, which was previously thought to be impossible. The purpose of this research is to develop, for the benefit of the deaf community, a system that can translate speech into text. Because it’s possible that some individuals won’t comprehend only the written word, the speech will also be translated into international sign language (ISL). In this method, persons who are deaf or hard of hearing will have their sign language translated into spoken language. In order to develop the most accurate model possible, we will integrate NLP with a variety of ML and AI techniques. For the purpose of prediction, convolutional neural networks (CNN) will be used since lip movements are continuous and rapid, making them challenging to capture. CNN and attention-based long and short-term memory (LSTM) approaches will function since they are both effective at predicting visual input. In order to get more desirable results, we will be using several techniques for augmenting the data. TensorFlow and Keras are going to be the Python libraries that are employed in order to do the voice-to-text translation. At the moment, there is an abundance of software accessible;however, all of them require the use of a network. This piece of hardware can function very well without being connected to the internet. Using the suggested technique, we were able to attain a 100% accuracy rate in sign language prediction and a 96% a

关键词： Forecasting

来源：评论

学校读者我要写书评

暂无评论

RSM-NLP at BLP-2023 Task 2: Bangla Sentiment Analysis using Weighted and Majority Voted Fine-Tuned Transformers 1

RSM-NLP at BLP-2023 Task 2: Bangla Sentiment Analysis using ...

引用

1st Workshop on Bangla Language Processing, BLP 2023

作者： Seth, Pratinav Goel, Rashi Mathur, Komal Vemulapalli, Swetha Dept. of Data Science & Computer Applications Manipal Institute of Technology Manipal Academy of Higher Education Manipal India Dept. of Computer Science & Engineering Dept. of Information and Communication Technology

ISBN: (纸本)9798891760585

This paper describes our approach to submissions made at Shared Task 2 at BLP Workshop - Sentiment Analysis of Bangla Social Media Posts(Hasan et al., 2023a;Islam et al., 2021;Hasan et al., 2023b). Sentiment Analysis is an action research area in the digital age. With the rapid and constant growth of online social media sites and services and the increasing amount of textual data, the application of automatic Sentiment Analysis is on the rise. However, most of the research in this domain is based on the English language. Despite being the world’s sixth most widely spoken language, little work has been done in Bangla. This task aims to promote work on Bangla Sentiment Analysis while identifying the polarity of social media content by determining whether the sentiment expressed in the text is Positive, Negative, or Neutral. Our approach consists of experimenting and finetuning various multilingual and pre-trained BERT-based models on our downstream tasks and using a Majority Voting and Weighted ensemble model that outperforms individual baseline model scores. Our system scored 0.711 for the multiclass classification task and scored 10th place among the participants on the leaderboard for the shared task. Our code is available at https://***/ptnv-s/RSM-NLP-BLP-Task2. © 2023 Association for Computational Linguistics.

关键词： Sentiment analysis

来源：评论

学校读者我要写书评

暂无评论

没有更多数据了...

全选清除本页清除全部题录导出标记到“检索档案”

共362页 << < 27 28 29 30 31 32 33 34 35 36 > >>

检索报告对象比较合并检索0

隐藏清空

合并搜索

回到顶部

执行限定条件

内容：

评分：

请选择保存的检索档案：

请选择收藏分类：

订阅名称：

通借通还

温馨提示：

图书名称：

借书校区：

取书校区：

手机号码：

邮箱地址：

一卡通帐号：

电话和邮箱必须正确填写，我们会与您联系确认。

联系人：

所在院系：

联系邮箱：

联系电话：

内蒙古自治区呼和浩特市赛罕区大学西街235号邮编: 010021

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：