检索结果-内蒙古大学图书馆

arXiv 2024年

作者： Zhang, Zhi Yadav, Srishti Han, Fengze Shutova, Ekaterina ILLC University of Amsterdam Netherlands Dept. of Computer Science University of Copenhagen Denmark Dept. of Computer Engineering Technical University of Munich Germany

The recent advancements in auto-regressive multimodal large language models (MLLMs) have demonstrated promising progress for vision-language tasks. While there exists a variety of studies investigating the processing of linguistic information within large language models, little is currently known about the inner working mechanism of MLLMs and how linguistic and visual information interact within these models. In this study, we aim to fill this gap by examining the information flow between different modalities—language and vision—in MLLMs, focusing on visual question answering. Specifically, given an image-question pair as input, we investigate where in the model and how the visual and linguistic information are combined to generate the final prediction. Conducting experiments with a series of models from the LLaVA series, we find that there are two distinct stages in the process of integration of the two modalities. In the lower layers, the model first transfers the more general visual features of the whole image into the representations of (linguistic) question tokens. In the middle layers, it once again transfers visual information about specific objects relevant to the question to the respective token positions of the question. Finally, in the higher layers, the resulting multimodal representation is propagated to the last position of the input sequence for the final prediction. Overall, our findings provide a new and comprehensive perspective on the spatial and functional aspects of image and language processing in the MLLMs, thereby facilitating future research into multimodal information localization and editing. Copyright © 2024, The Authors. All rights reserved.

关键词： Visual languages

来源：评论

学校读者我要写书评

暂无评论

Air Pollution Forecasting Using Machine Learning with Temporal Fusion Transformer and Graph Neural Networks

Air Pollution Forecasting Using Machine Learning with Tempor...

引用

Intelligent Data Communication Technologies and Internet of Things (IDCIoT), International Conference on

作者： Subhashree Rath Maneesha P dept. of Computer Science and Engineering New Horizon College Of Engineering Autonomous College Affiliated to VTU Bengaluru

ISBN: (数字)9798331527549

ISBN: (纸本)9798331527556

Air pollution is a significant obstacle to achieving urban sustainability and public health, particularly in rapidly developing regions. Accurate air quality prediction is essential for proactive pollution management and policy formulation. In this study, we employ sophisticated machine learning frameworks, such as Temporal Fusion Transformer (TFT) and Graph Neural Networks (GNNs), to predict air quality levels. TFT is designed to capture temporal dependencies, making it ideal for time-series forecasting, while GNNs model spatial relationships, enabling the analysis of pollutant dispersion across monitoring stations. A synthetic dataset comprising 50 rows was generated, simulating pollutant concentrations (PM2.5, PM10, CO, NO2, SO2) and meteorological variables (temperature, humidity, wind speed). The models were assessed using criteria like $R^{2}$ , RMSE, and MAE. TFT achieved an $R^{2}$ of 0.97 and an RMSE of $4.2 \mu ~\mathrm{g} / \mathrm{m}^3$ , significantly outperforming traditional models like Random Forest and XGBoost. GNN provided spatial insights, identifying pollutant hotspots and their dispersion patterns. This research highlights the superior accuracy and scalability of TFT and GNN for air quality prediction. The results underline their potential for real-time monitoring, smart city integration, and proactive environmental management, offering a robust solution for addressing urban air pollution challenges.

关键词： Accuracy Atmospheric modeling Predictive models Transformers Air pollution Graph neural networks Real-time systems Thin film transistors Forecasting Dispersion

来源：评论

学校读者我要写书评

暂无评论

Real-World Atmospheric Turbulence Correction Via Domain Adaptation

Real-World Atmospheric Turbulence Correction Via Domain Adap...

引用

IEEE International Conference on Image Processing

作者： Xijun Wang Santiago López-Tapia Aggelos K. Katsaggelos Dept. of Computer Science Northwestern University Evanston IL USA Dept. of Electrical and Computer Engineering Northwestern University Evanston IL USA

ISBN: (数字)9798350349399

ISBN: (纸本)9798350349405

Atmospheric turbulence, a common phenomenon in daily life, is primarily caused by the uneven heating of the Earth’s surface. This phenomenon results in distorted and blurred acquired images or videos and can significantly impact downstream vision tasks, particularly those that rely on capturing clear, stable images or videos from outdoor environments, such as accurately detecting or recognizing objects. Therefore, people have proposed ways to simulate atmospheric turbulence and designed effective deep learning-based methods to remove the atmospheric turbulence effect. However, these synthesized turbulent images can not cover all the range of real-world turbulence effects. Though the models have achieved great performance for synthetic scenarios, there always exists a performance drop when applied to real-world cases. Moreover, reducing real-world turbulence is a more challenging task as there are no clean ground truth counter-parts provided to the models during training. In this paper, we propose a real-world atmospheric turbulence mitigation model under a domain adaptation framework, which links the supervised simulated atmospheric turbulence correction with the unsupervised real-world atmospheric turbulence correction. We will show our proposed method enhances performance in real-world atmospheric turbulence scenarios, improving both image quality and downstream vision tasks.

关键词： Image quality Training Learning systems Adaptation models Atmospheric modeling Prevention and mitigation Image restoration

来源：评论

学校读者我要写书评

暂无评论

PV-Power Forecasting using Machine Learning Techniques

PV-Power Forecasting using Machine Learning Techniques

引用

IEEE International Conference on Electro-Information Technology

作者： Kazi Abdullah Al Arafat Kode Creer Anjan Debnath Temitayo O. Olowu Imtiaz Parvez Dept. of Computer Sci and Eng Atish Dipankar University of Science and Technology Dhaka Bangladesh Dept. of Computer Science Utah Valley University Orem UT Dept. of Electrical Engineering and Computer Science OR Idaho National Laboratory Idaho Falls ID

ISBN: (数字)9798350330649

ISBN: (纸本)9798350330656

Solar energy forecasting plays a pivotal role in the efficient utilization of renewable energy resources for sustainable power generation. This study delves into the domain of solar-power forecasting, employing a comprehensive analysis of machine learning models. The primary objective is to evaluate and compare the performance of Gated Recurrent Unit (GRU), Recurrent Neural Network (RNN), Multi-Layer Perceptron (MLP), and Linear Regression (LR) models in predicting solar energy production. Through a comprehensive evaluation of individual model performance, the study provides nuanced insights into the strengths and limitations of each forecasting approach. Results indicate that the Multy-Layer Perceptron (MLP) model excels in accuracy, exhibiting low root mean square error (RMSE) and high correlation among the parameters. The Gated Recurrent Unit (GRU) model demonstrates competitive performance, while the Recurrent Neural Network model showcases strengths in multiple metrics. Additionally, MLP and GRU models display superior predictive accuracy, emphasizing their efficacy in solar energy forecasting.

关键词： Renewable energy sources Recurrent neural networks Correlation Accuracy Machine learning algorithms Solar energy Machine learning

来源：评论

学校读者我要写书评

暂无评论

Random Selection for Cloud-Based Load-Distributing in Varied P2P Networks using Enhanced Throttled Algorithm 5

Random Selection for Cloud-Based Load-Distributing in Varied...

引用

5th IEEE International Conference on Communication, Computing and Industry 6.0, C2I6 2024

作者： Prakash, V.S. Mary, Margaret T. Shivakumar, B.N. Gnanadurai, Jasmine Divya, K.S. Vennila, J. Kristu Jayanti College Department of Computer Science Karnataka Bangalore India Cmr Institute of Technology Department of Mathematics Karnataka Bengaluru India George Fox University Dept. Electrical and Computer Science Engineering OR United States Manipal College of Health Professions Mahe Department of Statistics Manipal India

ISBN: (纸本)9798350379884

A peer-to-peer network connection is essential to most computer users' daily activities. On heterogeneous P2P networks, maintaining cloud-based load balance is a difficult goal to achieve. Several previous research have provided load rearrangement procedures for additional load diversions, and load division on the point where the original peer arrives. The goal of cloud-based load-sharing in P2P structures is to evenly distribute the burden among net nodes, then it is not effective in efficiently removing traffic that is already happening on the P2P network's routes. By removing the traffic flow and consuming the Random Selection Load Balancing (RSLB) method and the Incredible Store Model, the behaviors in this study are characterized. Using an improved throttled method, the Incredible Store Model is examined to broaden the effects of node variety in addition to agitation toward the capacity circulation in P2P systems. Now different from the current load distribution balancing strategy, the presentation of the Random Selection Load Balancing (RSLB) technique by the Incredible Store Model decreases the network traffic flow pathways with little cloud rate of P2P. By combining the burden in the P2P net system, an investigative and observed conclusion delivers a destruction-free system with information calculated. Peer failure probability in addition to traffic control effectiveness is used to estimate the presentation of the random selection load assessment. © 2024 IEEE.

关键词： Peer to peer networks

来源：评论

学校读者我要写书评

暂无评论

Pattern Recognition in Lung Disease Classification: Leveraging Convolutional Neural Networks that Employs Local Phase Quantization Features

Pattern Recognition in Lung Disease Classification: Leveragi...

引用

International Conference on Computing and Networking Technology (ICCNT)

作者： M Ramya Sree Mettu Siddhartha Poli Vamsi Vardhan Reddy Tripty Singh Rekha R Nair Dept. of Computer Science and Engineering Amrita School of Computing Amrita Vishwa Vidyapeetham Bengaluru India Dept. of Computer Science and Engineering Alliance School of Advanced Computing Alliance University Bangalore India

ISBN: (数字)9798350370249

ISBN: (纸本)9798350370270

Lung diseases are wide group of diseases related to the respiratory system, and is ranging from acute infections to chronic disorders with multifactorial causes. Recognition of the lung disease is very crucial in the proper diagnosis, forecasting through prognosis, and remedies. Hence proposed lung disease classification with feature extraction technique. The proposed work carried out feature extraction with Gray-Level Co-occurrence Matrix (GLCM) and Local Phase Quantization (LPQ) combined with Convolutional Neural Network(CNN). The goal is to create a model that reduces diagnostic errors, ensuring more precise early detection and classification of pulmonary disorders. This initiative facilitates the advancement of pulmonary medicine by promoting accurate radiologic imaging, while future directions emphasize integrating medical workflows to enable personalized and timely diagnosis and treatment. Additionally, the ResNet model demonstrates superior performance in hernia detection, achieving a test accuracy of 89.43% for GLCM based feature extraction and Pneumonia detection with 79.20% test accuracy for LPQ based feature extraction.

关键词： Visualization Accuracy Quantization (signal) Pneumonia Lungs Computational modeling Feature extraction Convolutional neural networks X-ray imaging Medical diagnostic imaging

来源：评论

学校读者我要写书评

暂无评论

Short-term Wind Power Ramp Forecasting Using Sequential Approach

Short-term Wind Power Ramp Forecasting Using Sequential Appr...

引用

Modeling, Simulation & Intelligent Computing (MoSICom), International Conference on

作者： Leechita Gopalakrishnan Julian L Cardenas Barrera N. Sabiyath Fatima Dept. of Computer Science and Engineering B.S. Abdur Rahman Crescent Institute of Science and Technology Chennai India Dept. of Electrical and Computer Engineering University of New Brunswick Fredericton Canada

ISBN: (数字)9798331533311

ISBN: (纸本)9798331533328

As wind energy becomes a key player in the global transition to renewable energy, its variability presents significant challenges. Accurate wind ramp forecasting is crucial for managing such rapid fluctuations in wind power output, ensuring grid stability, and enhancing the reliability of renewable energy integration. This paper presents a short-term wind ramp forecasting model using a sequential approach: initially forecasting wind power through the integration of Variational Mode Decomposition (VMD) and the XGBoost algorithm, and subsequently detecting wind ramps using the Definition Based Sign Indicator (DSI) method. The historical wind speed and power data for Turkey was obtained from Kaggle. The dataset comprises readings from a wind turbine’s SCADA system in Turkey, with measurements recorded at 10minute intervals. Variational Mode Decomposition (VMD) was applied to deconstruct and restore the historical wind power of the wind field, resulting in the formation of three training datasets. Wind power projections for the subsequent 8 hours, at 10 -minute intervals, are generated using historical wind power, wind speed, and wind direction as inputs. The DSI algorithm is employed to identify ramps. To assess the predictive capability of the proposed model, two algorithms, SVR and XGBoost, were employed for forecasting and the resulting error is minimal. The proposed model, VMD + XGBOOST is also compared without VMD to evaluate the performance. Taking mean square error (MSE), root mean square error (RMSE) mean absolute error (MAE), and symmetric mean absolute percentage error (sMAPE) as evaluation indicators, the results show that the forecasted accuracy of wind power is significantly improved after VMD processing. The performance of XGBOOST is better than SVR in the two evaluation indicators. Precision, Recall, and F1-score are calculated to evaluate the accuracy of the predicted wind ramps.

关键词： Renewable energy sources Accuracy Wind speed SCADA systems Wind power generation Predictive models Prediction algorithms Wind turbines Wind forecasting Forecasting

来源：评论

学校读者我要写书评

暂无评论

Automatic Feature Learning for Essence: a Case Study on Car Sequencing

arXiv

引用

arXiv 2024年

作者： Pellegrino, Alessio Akgün, Özgür Dang, Nguyen Kiziltan, Zeynep Miguel, Ian Dept. of Computer Science and Engineering University of Bologna Italy School of Computer Science University of St Andrews United Kingdom

Constraint modelling languages such as Essence offer a means to describe combinatorial problems at a high-level, i.e., without committing to detailed modelling decisions for a particular solver or solving paradigm. Given a problem description written in Essence, there are multiple ways to translate it to a low-level constraint model. Choosing the right combination of a low-level constraint model and a target constraint solver can have significant impact on the effectiveness of the solving process. Furthermore, the choice of the best combination of constraint model and solver can be instance-dependent, i.e., there may not exist a single combination that works best for all instances of the same problem. In this paper, we consider the task of building machine learning models to automatically select the best combination for a problem instance. A critical part of the learning process is to define instance features, which serve as input to the selection model. Our contribution is automatic learning of instance features directly from the high-level representation of a problem instance using a language model. We evaluate the performance of our approach using the Essence modelling language with a case study involving the car sequencing problem. © 2024, CC BY.

关键词： Modeling languages

来源：评论

学校读者我要写书评

暂无评论

Towards Robust mmWave-based Human Activity Recognition using Large Simulated Dataset for Model Pretraining

Towards Robust mmWave-based Human Activity Recognition using...

引用

IEEE International Conference on Big Data

作者： Vinay Joshi Shengkai Xu Qiming Cao Yi Zhu Pu Wang Hongfei Xue Dept. of Computer Science UNC Charlotte Charlotte USA Electrical and Computer Engineering Purdue University West Lafayette USA Dept. of Computer Science Wayne State University Detroit USA

ISBN: (数字)9798350362480

ISBN: (纸本)9798350362497

Human activity recognition (HAR) is crucial for real-world applications such as healthcare, surveillance, and smart homes. Among sensing technologies, millimeter wave (mmWave) sensors stand out due to their contactless nature, high sensitivity, and ability to operate in low-light environments while preserving privacy. However, the scarcity of mmWave sensing data limits the generalizability of mmWave-based HAR systems. To address this, we propose mmAP, a data augmentation and pretraining framework that synthesizes a large mmWave dataset using human mesh data, followed by pretraining a robust and general mmWave heatmap encoder using a multi-modal masked autoencoder framework using the synthesized data. We enhance the model’s robustness with heatmap-specific data perturbations and perform task-specific fine-tuning on a small real-world dataset. The experiment results over the baseline demonstrate the effectiveness of the proposed mmAP framework.

关键词： Heating systems Perturbation methods Surveillance Autoencoders Smart homes Data augmentation Data models Robustness Human activity recognition Millimeter wave communication

来源：评论

学校读者我要写书评

暂无评论

Vertical validation: evaluating implicit generative models for graphs on thin support regions 24

Vertical validation: evaluating implicit generative models f...

引用

Proceedings of the Fortieth Conference on Uncertainty in Artificial Intelligence

作者： Mai Elkady Thu Bui Bruno Ribeiro David I. Inouye Computer Science Dept. Purdue University West Lafayette Indiana Electrical and Computer Engineering Dept. Purdue University West Lafayette Indiana

There has been a growing excitement that implicit graph generative models could be used to design or discover new molecules for medicine or material design. Because these molecules have not been discovered, they naturally lie in unexplored or scarcely supported regions of the distribution of known molecules. However, prior evaluation methods for implicit graph generative models have focused on validating statistics computed from the thick support (e.g., mean and variance of a graph property). Therefore, there is a mismatch between the goal of generating novel graphs and the evaluation methods. To address this evaluation gap, we design a novel evaluation method called Vertical Validation (VV) that systematically creates thin support regions during the train-test splitting procedure and then reweights generated samples so that they can be compared to the held-out test data. This procedure can be seen as a generalization of the standard train-test procedure except that the splits are dependent on sample features. We demonstrate that our method can be used to perform model selection if performance on thin support regions is the desired goal. As a side benefit, we also show that our approach can better detect overfitting as exemplified by memorization.

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：