检索结果-内蒙古大学图书馆

2022 IEEE North Karnataka Subsection Flagship International Conference, NKCon 2022

作者： Nithyapriya, S. Keerthika, J. Tamizh Kuzhali, R. Saranya, M.D. Revathi, P. Daniya, T. Bannari Amman Institute of Technology Department of Artificial Intelligence and Data Science Sathyamangalam India Sri Eshwar College of Engineering Department of Computer Science and Engineering Coimbatore India Sree Sastha Institution of Engineering and Technology Department of Computer Science and Engineering Chennai India KPR Institute of Engineering and Technology Department of Electronics and Communication Engineering Coimbatore India Hindusthan College of Engineering and Technology Department of Information Technology Coimbatore India GMR Institute of Technology Department of Information Technology Andhra Pradesh Rajam India

ISBN: (纸本)9781665453424

Pneumonia is a respiratory infection caused by bacteria, virus or fungi. This infection affects the lungs by filling the air sacs(alveoli) with fluid or pus. Chest X-ray(CXR) imaging is the preferred diagnostic examination to locate the inflammation in the lungs. Examining the chest X-rays with high diagnostic accuracy is very difficult for the medical practitioners. To counteract this, computer-aided diagnosis system is used in this study for classifying the Chest X-ray (CXR) images into pneumonia and normal lungs. This research uses transfer learning based stacked ensemble CNN snapshots for the pneumonia prediction from the chest X-ray images. The DenseNet-121 architecture is used as the base learner. To reduce the computational time, the model is trained once to yield several CNN snapshots. The created snapshots give various details regarding the features taken from the CXR. The decision scores of multiple snapshots are combined using the stacking ensemble technique through a metalearner(Random Forest). The proposed ensemble model obtains an accuracy score of 98.36% on the publicly available Kermany dataset. © 2022 IEEE.

关键词： Air

来源：评论

学校读者我要写书评

暂无评论

Ranking on Heterogeneous Manifold for Multimodal Information Retrieval 21

Ranking on Heterogeneous Manifold for Multimodal Information...

引用

21st IEEE International Symposium on Parallel and Distributed Processing with Applications, 13th IEEE International Conference on Big data and Cloud Computing, 16th IEEE International Conference on Social Computing and Networking and 13th International Conference on Sustainable Computing and Communications, ISPA/BDCloud/SocialCom/SustainCom 2023

作者： Jin, Zhenkun Wan, Xingshi Nie, Xin Zhou, Xinlei Yi, Yuanyuan Zhou, Gefei Wuhan Business University Wuhan Research Institute of Posts and Telecommunications Wuhan China Wuhan Research Institute of Posts and Telecommunications Wuhan China Huazhong University of Science and Technology Hubei Key Laboratory of Distributed System Security Hubei Engineering Research Center on Big Data Security Wuhan China Huazhong University of Science and Technology School of Journalism and Information Communication Wuhan China Huazhong University of Science and Technology School of Computer Science & Technology Wuhan China

ISBN: (纸本)9798350329223

In light of recent advancements in Internet of Multimedia Things (IoMT) and 5G technology, both the variety and quantity of data have been rapidly increasing. Consequently, handling zero-shot cross-modal retrieval (ZSCMR) of unlabeled multi-modal data has emerged as a prominent research topic. Early models utilized generative adversarial networks and class embeddings to assist in building a common feature space. However, this approach resulted in a loss of reliability in the feature distributions of individual modalities. In this paper, we propose a method called Ranking on Heterogeneous Manifold (RHM) to address the aforementioned issue. Our model comprises three feature extractors, each dedicated to extracting features from text, images and multimodal data. Subsequently, the relationship structures derived from these feature extractors are employed to construct a heterogeneous manifold. A diffusion process is then conducted on this heterogeneous manifold, ultimately yielding the final ranking scores. Comprehensive experiments demonstrate that the proposed modal has the capability to acquire more effective knowledge representation, thus enhancing the effectiveness of ZSCMR. © 2023 IEEE.

关键词： Knowledge representation

来源：评论

学校读者我要写书评

暂无评论

Invisible Watermark and Hypothesis Testing for Multi-Ownership Verification Framework of Deep Neural Network

Invisible Watermark and Hypothesis Testing for Multi-Ownersh...

引用

IEEE International Symposium on Information (IT) in Medicine and Education, ITME

作者： Ruoxi Wang Yujia Zhu Daoxun Xia School of Big Data and Computer Science Guizhou Normal University Key Laboratory of Information and Computing Science Guizhou Province Guizhou Normal University Engineering Laboratory for Applied Technology of Big Data in Education Guizhou Normal University Guiyang China

ISBN: (数字)9798350356670

ISBN: (纸本)9798350356687

With the rapid development of information technology, artificial intelligence models have become the core asset of tech giants, highlighting the importance of copyright protection. Black-box watermarking has gained attention due to its ability to verify the legality of models remotely without relying on internal model information. The challenges lie in embedding and verifying watermarks without accessing internal model parameters, while ensuring the robustness and stealth of the watermark. This paper focuses on discussing effective ways to protect the copyright of black-box watermarking models and detect and prevent unauthorized user behavior that infringes on model ownership. Therefore, this paper proposes a multi-ownership verification framework based on invisible watermark and hypothesis testing, which mainly consists of two parts: watermark generation and copyright verification. The watermark generation uses a dynamic watermarking method to generate invisible backdoor watermark and verifies copyright through hypothesis testing. This framework has high assurance and effectiveness under various models of CIFAR-10 and MNIST datasets and shows strong robustness in VGG and Resnet model pruning attacks. This research provides new ideas and methods for the protection of model copyrights in deep neural networks.

关键词： Deep learning Education Closed box Watermarking Artificial neural networks Copyright protection Benchmark testing Robustness Protection Information technology

来源：评论

学校读者我要写书评

暂无评论

Symmetrical Supervision with Transformer for Few-shot Medical Image Segmentation

Symmetrical Supervision with Transformer for Few-shot Medica...

引用

2022 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2022

作者： Niu, Yao Luo, Zhiming Lian, Sheng Li, Lei Li, Shaozi Song, Haixin Xiamen University Department of Artificial Intelligence Fujian China Fuzhou University College of Computer and Data Science Fujian China Xiamen University Department of Software Engineering Fujian China

ISBN: (纸本)9781665468190

Few-shot learning can potentially learn the target knowledge in extremely few data regimes. Existing few-shot medical image segmentation methods fail to consider the global anatomy correlation between the support and query sets. They generally adopt a weak one-way information transmission that can not fully explore the knowledge to segment query data. To address this problem, we propose a novel Symmetrical Supervision network based on traditional two-branch methods. We raise two main contributions: (1) The Symmetrical Supervision Mechanism is leveraged to strengthen the supervision of network training;(2) A transformer-based Global Feature Alignment module is introduced to increase the global consistency between the two branches. Experimental results on two challenging datasets (abdominal segmentation dataset CHAOS and cardiac segmentation dataset MS-CMRSeg) show a remarkable performance compared to other comparing methods. © 2022 IEEE.

关键词： Image segmentation

来源：评论

学校读者我要写书评

暂无评论

Revolutionizing Cryptocurrency Price Prediction: Advanced Insights from Machine Learning, Deep Learning and Hybrid Models

Revolutionizing Cryptocurrency Price Prediction: Advanced In...

引用

Future of Information and Communication Conference, FICC 2025

作者： Amberkhani, Abdullah Bolisetty, Harshitha Narasimhaiah, Ranjith Jilani, Ghulam Baheri, Betis Muhajab, Hanan Muhajab, Areej Ghazinour, Kambiz Shubbar, Safa Department of Data Science Kent State University KentOH44242 United States Department of Computer Science Kent State University KentOH44242 United States Department of Computer Science College of Engineering and Computer Science Jazan University Jazan Saudi Arabia Computers and Information Technology College Taif University Taif Saudi Arabia CantonNY United States

ISBN: (纸本)9783031844591

As of January 2024, the global cryptocurrency market cap reached $1.66 trillion, marking a 1.5% increase from December 2023. Cryptocurrencies, driven by cryptographic algorithms like SHA-2 and MD5, offer a volatile yet promising investment opportunity. The unpredictable nature of this market has led to the development of various prediction models—such as GRUs, LSTMs, ARIMAX, and others—aimed at forecasting price movements. This study explores a range of machine learning and deep learning algorithms, including SVR, RFR, XGBoost, RNN, CNN, LSTM, the LSTM-GRU hybrid, and fbProphet, to predict cryptocurrency prices with enhanced accuracy. While models like XG- Boost and the LSTM-GRU hybrid demonstrated robust performance, this research also highlights the challenges of overfitting and the importance of data preprocessing and model tuning in improving prediction reliability. The findings provide valuable insights for investors seeking precise forecasts in the dynamic cryptocurrency market. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

关键词： Deep learning

来源：评论

学校读者我要写书评

暂无评论

Avoiding exp(Rmax) scaling in RLHF through Preference-based Exploration

arXiv

引用

arXiv 2025年

作者： Chen, Mingyu Chen, Yiding Sun, Wen Zhang, Xuezhou Department of Electrical & Computer Engineering Boston University United States Department of Computer Science Cornell University United States Faculty of Computing & Data Sciences Boston University United States

Reinforcement Learning from Human Feedback (RLHF) has emerged as a pivotal technique for large language model (LLM) alignment. This paper studies the setting of online RLHF and focus on improving sample efficiency. All existing algorithms in online RLHF, whether doing passive exploration or active exploration, suffer from a sample complexity that scales exponentially with the scale of the reward function. This fundamental limitation hinders their effectiveness in scenarios with heavily skewed preferences, e.g. questions with a unique correct solution. To address this, we introduce Self-Exploring Preference-Incentive Online Preference Optimization (SE-POPO), an online RLHF algorithm that for the first time achieves a sample complexity that scales polynomially with the reward scale, answering an open problem raised by Xie et al. (2024). Theoretically, we demonstrate that the sample complexity of SE-POPO dominates that of existing exploration algorithms. Empirically, our systematic evaluation confirms that SE-POPO is more sample-efficient than both exploratory and non-exploratory baselines, in two primary application scenarios of RLHF as well as on public benchmarks, marking a significant step forward in RLHF algorithm design. The code is available at https://***/MYC000801/SE-POPO. © 2025, CC BY.

关键词： Reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

Knowledge Enhanced Coreference Resolution via Gated Attention

Knowledge Enhanced Coreference Resolution via Gated Attentio...

引用

2022 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2022

作者： He, Kai Mao, Bing Zhou, Xiangyu Li, Yufei Gong, Tieliang Li, Chen Wu, Jialun Xi'an Jiaotong University School of Computer Science and Technology Xi'an China Xi'an Jiaotong University Shaanxi Provincial Key Laboratory of Big Data Knowledge Engineering Xi'an China

ISBN: (纸本)9781665468190

Coreference resolution aims at linking all mentions that refer to the same entity, which are widely adopted in many biomedical and bioinformatics tasks, such as biomedical knowledge graph construction and metabolic pathway integration. Many recent studies focus on improving neural model structures. However, we argue that a practical method that integrates commonsense knowledge can further improve coreference resolution performance, because commonsense delivers extra prior knowledge for reasoning and can enhance related representations, rather than naive mention-context occurrence modeling. In this work, we propose an effective method to integrate external commonsense knowledge into a neural coreference resolution model. Specially, a gated attention mechanism is employed in our method to leverage commonsense according to different contexts. By using ConceptNet as the knowledge base in three span-ranking backbone models, the models can yield significant performance gains on used datasets. We also achieve improvements in tasks of long-term mention detection and cross-sentence coreferences after incorporating knowledge. © 2022 IEEE.

关键词： Model structures

来源：评论

学校读者我要写书评

暂无评论

A Design of an Effective Methodology to Detect Glaucoma Disease using Intelligent Learning Strategy 9

A Design of an Effective Methodology to Detect Glaucoma Dise...

引用

9th IEEE International Conference on Smart Structures and Systems, ICSSS 2023

作者： Shobana, D. Akilarasu, G. Sangeetha, J. Priya, V. Banu Karkuzhali, S. Malathi, P. Rajalakshmi Engineering College Department of Electrical and Electronics Engineering Tamil Nadu Chennai India Lovely Professional University Department of Computer Science and Engineering Punjab India S. A. Engineering College Department of Computer Science and Engineering Tamil Nadu Chennai India R.M.K. College of Engineering and Technology Department of Mathematics Tamil Nadu Chennai India R.M.K. Engineering College Department of General Engineering Tamil Nadu Chennai India Saveetha College of Liberal Arts and Sciences Simats Department of Data Analytics Tamil Nadu Chennai India

ISBN: (纸本)9798350384208

Globally, glaucoma is now among the leading causes of blindness. It is a chronic, irreversible illness. Screening for glaucoma and detecting it early are crucial in the fight against this illness. Here, deep learning shows great promise. Two deep learning architectures, the Convolutional Neural Network (CNN) model and the Improved Learning Strategy to Predict Glaucoma (ILSPG) model, have been evaluated in this article. In order to diagnose glaucoma, the DCNN method is utilized. Ultimately, this area of study aims to help doctors diagnose glaucoma by developing methods to spot the disease in retinal fundus pictures. Because glaucoma can have both beneficial and bad effects on the model, the result of the model might be good or bad. Several measures are utilized in the process of evaluating models, including accuracy, precision, specificity, and reliability. In order to determine the accuracy of the proposed model, an additional comparison study is conducted. The results are contrasted with a deep learning-based classification system that uses convolutional neural networks. The results demonstrate, without a doubt, that the proposed system is effective. © 2023 IEEE.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Machine Learning and Multiple Regression-Based Study of the Medicinal Properties of Food

Machine Learning and Multiple Regression-Based Study of the ...

引用

作者： Jiasheng Zhou Zhihong Fan Wenxin Zhang Fuzhou University College of Computer and Data Science/College of Software Fuzhou University College of Physics and Information Engineering

We study the relationship between food ingredients and hot and cold properties based on the idea that "medicine and food have the same origin". Firstly, we classify foods with known hot and cold properties as typical representatives of flat, warm and cold properties, and use various machine learning algorithms to classify the typical food representatives. In order to further improve the reliability and accuracy of the classification, we applied Bayesian optimization to the SVM and the SVM was able to classify the typical food representatives again with an accuracy of 96.53%. Based on the above findings, we further analysed which chemical components played a key role in the hot and cold properties. We then applied multivariate logistic regression for quantitative analysis, using stepwise forward regression to minimise complete multicollinearity and OLS + robust standard errors to eliminate the effect of heteroscedasticity. Based on the analysis of the coefficients of the regression equations and the significance test results, the conclusions reached were consistent with the qualitative analysis, with the four main components of energy, water, minerals and fat playing a major role in the chilling and heating properties of food. In conjunction with the analysis in the full text, we make recommendations for the development of functional foods where heat and cold are the principles.

关键词： Factor analysis Clustering algorithms Food classification

来源：评论

学校读者我要写书评

暂无评论

From data to Harvest through Machine Learning based Crop Yield Forecasting

From Data to Harvest through Machine Learning based Crop Yie...

引用

Advancements in Smart, Secure and Intelligent Computing (ASSIC), International Conference on

作者： Aadithya Vikram Budarapu John Hyde Gaddam Ramu Boyedi Thoshan Kumer Muthyala Nikhila Kathirisetty Dept. of Information Technology Vardhaman College of Engineering (Affiliated by JNTUH) Hyderabad India Dept. of Computer Science and Data Science Vardhaman College of Engineerin (Affiliated by JNTUH) Hyderabad India

The maintenance and enhancement of dynamic soil characteristics are the primary focus of soil management in agriculture to increase crop productivity. Higher productivity may result from efficient soil control of resources and corrective micronutrient treatments. Using CNN and ’KNN’ algorithms, the “soil land classification and crop prediction system” application was created. In this study, two datasets are used: one to obtain crop prediction and the other for soil land categorization. The kind of soils are trained using CNN (“VGG- 19”) algorithm, and the accuracy of the model is calculated. The trained model is then utilized in the Flask web app to forecast the type of soil. Another data set with nitrogen, phosphorus, potassium, pH, and temperature as features and the class type of crop as a label is used to forecast crop production. These two algorithms are used to create the flask website, which accepts inputs such as soil picture, soil type prediction, and land parameter inputs for crop prediction.

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：