检索结果-内蒙古大学图书馆

Reconstructing Optical Imagery from Microwave data: A Neural Regression-Based Framework

学校读者我要写书评

暂无评论

Reconstructing Optical Imagery from Microwave Data: A Neural...

Remote Sensing Symposium (InGARSS), 2020 IEEE India Geoscience and

作者： Arpan Dawn Kounik De Sarkar Abhiroop Chatterjee Susmita Ghosh Surajit Ghosh Department of Civil Engineering National Institute of Technology Durgapur Durgapur India Environmental Science and Engineering Department Indian Institute of Technology Bombay Mumbai India Department of Computer Science and Engineering Jadavpur University Kolkata India Water Futures Data and Analytics International Water Management Institute Colombo Sri Lanka

ISBN: (数字)9798350390346

ISBN: (纸本)9798350390353

To monitor the vegetation condition during extreme weather events, this article suggests an approach for mimicking vegetation indices, notably the Normalized Difference Vegetation Index (NDVI), using microwave data. The main goal is to generate simulated NDVI images from microwave data to improve temporal availability and fill the data gaps in the time series when optical data is unavailable. A neural network regression model was developed, using SAR data as input features to predict optical data. The network architecture included an input layer, followed by a dense layer with 32 ReLU-activated nodes, and an output layer with a single node. The model was trained using the Adam optimizer and mean squared error (MSE) as the loss function. Initial results show an $R^{2}$ of 0.62. Pearson’s correlation coefficient is $\mathbf{- 0. 7 9}$ and 0.019 for red vs VV and NIR vs VH bands, respectively.

关键词： Normalized difference vegetation index Clouds Ultraviolet sources Vegetation mapping Optical imaging Cyclones Adaptive optics Optical sensors Monitoring Microwave imaging

A Novel Method to Identify and Recover the Fault Nodes over 5G Wireless Sensor Network Environment

学校读者我要写书评

暂无评论

A Novel Method to Identify and Recover the Fault Nodes over ...

2024 Asia Pacific Conference on Innovation in Technology, APCIT 2024

作者： Vijaya Vardan Reddy, S.P. Juliet, A. Hency Jayadurga, R. Sethu, S. Pavithra, K.N. Samuthira Pandi, V. R.M.K. Engineering College Department of Electronics and Communication Engineering Tamil Nadu Chennai India Saveetha Institute of Medical and Technical Sciences Department of Data Analytics Saveetha College of Liberal Arts and Sciences Saveetha University Tamil Nadu Chennai India Prince Shri Venkateshwara Padmavathy Engineering College Department of Master of Business Administration Tamil Nadu Chennai India Department of Computer Science and Engineering Tamilnadu Chennai India R. M. D. Engineering College Department of Science and Humanities Tamil Nadu Chennai India Chennai Institute of Technology Centre for Advanced Wireless Integrated Technology Tamil Nadu Chennai India

ISBN: (纸本)9798350361537

The significance of communication networks is growing in tandem with the proliferation of communication technologies. Present methods for maintaining 5G Wireless Sensor Networks are still restricted to routine maintenance and post-maintenance tasks. They lack a comprehensive function for monitoring the network's status, are unable to assess the network's health, and are difficult to maintain before the 5G-based WSNs seriously degrade. 5G Wireless Sensor Network faults can only be resolved by highly trained technicians due to low maintenance efficiency. As a result, errors cannot be detected or located promptly or accurately, leading to forced repairs that incur the expense of new network cables. First, the article lays forth the basics of network fault analysis. Then, it uses deep learning to simulate communication network problem diagnosis. Lastly, the experimental section compares various methodologies and analyses the findings of fault location. The results of the simulation demonstrate that the suggested approach mitigates the created model's flaws to a certain degree while simultaneously enhancing the network fault detection model's accuracy, universality, and robustness. A novel approach to autonomous placement, the Fault Node Recovery Protocol is described in this study. It is implemented in 5G mobile communications to detect faulty data according to wireless sensor network standards, and it is made to fix the problems with conventional techniques, such poor positioning precision and lengthy running time. The automated localization model for 5G mobile communication fault data, constructed using the suggested FNRP approach, is presented in this work. By comparing it to the standard Adhoc On-Demand Distance Vector Routing protocol, we can see how well the suggested system performs. © 2024 IEEE.

关键词： 5G mobile communication systems

Beyond Scaleup: Knowledge-aware Parsimony Learning from Deep Networks

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Yao, Quanming Zhang, Yongqi Wang, Yaqing Yin, Nan Kwok, James Yang, Qiang Department of Electronic Engineering the Tsinghua University China Department of Data Science and Analytics Thrust the Hong Kong University of Science and Technology Guangzhou China Beijing Institute of Mathematical Sciences and Applications China Department of Computer Science and Engineering the Hong Kong University of Science and Technology Hong Kong

The brute-force scaleup of training datasets, learnable parameters and computation power, has become a prevalent strategy for developing more robust learning models. However, due to bottlenecks in data, computation, and trust, the sustainability of this strategy is a serious concern. In this paper, we attempt to address this issue in a parsimonious manner (i.e., achieving greater potential with simpler models). The key is to drive models using domain-specific knowledge, such as symbols, logic, and formulas, instead of purely relying on scaleup. This approach allows us to build a framework that uses this knowledge as "building blocks" to achieve parsimony in model design, training, and interpretation. Empirical results show that our methods surpass those that typically follow the scaling law. We also demonstrate our framework in AI for science, specifically in the problem of drug-drug interaction prediction. We hope our research can foster more diverse technical roadmaps in the era of foundation models. © 2024, CC BY.

关键词： Contrastive Learning

Graph-based deep learning models in the prediction of early-stage Alzheimers

学校读者我要写书评

暂无评论

Graph-based deep learning models in the prediction of early-...

Annual International Conference of the IEEE engineering in Medicine and Biology Society (EMBC)

作者： Bishal Thapaliya Zundong Wu Ram Sapkota Bhaskar Ray Pranav Suresh Santosh Ghimire Vince Calhoun Jingyu Liu Department of Computer Science Georgia State University Atlanta USA Tri-Institutional Center for Translational Research in Neuroimaging and Data Science School of Electrical and Computer Engineering Georgia Institute of Technology Atlanta USA Department of Applied Sciences and Chemical Engineering Tribhuvan University Nepal

ISBN: (数字)9798350371499

ISBN: (纸本)9798350371505

Alzheimer's disease is the most common age-related problem and progresses in different stages, from cognitively normal to early mild cognitive impairment, and severe dementia. This study investigates the predictive potential of resting-state functional magnetic resonance imaging (rs-fMRI) and its derived functional connectivity (FC) in understanding Alzheimer's progression. Leveraging deep learning and graph-based models, we introduce two key contributions: 1) a comparative analysis of rs-fMRI time points and FC for Alzheimer's prediction. 2) an innovative graph transformer variant incorporating self-clustering for enhanced prediction accuracy. Experiments on the Alzheimer's Disease Neuroimaging Initiative dataset with 830 subjects reveal two notable conclusions. Firstly, rs-fMRI time points offer limited utility compared to functional network connectivity for transformer-based models, even when considering temporal information. Secondly, a clustering-based attention module proves effective for classifying brain networks in predicting Alzheimer's disease progression, providing valuable insights for future research and clinical applications.

关键词： Deep learning Neuroimaging Analytical models Biological system modeling Predictive models Functional magnetic resonance imaging Transformers Brain modeling Alzheimer's disease engineering in medicine and biology

Parm: Efficient Training of Large Sparsely-Activated Models with Dedicated Schedules

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Pan, Xinglin Lin, Wenxiang Shi, Shaohuai Chu, Xiaowen Sun, Weinong Li, Bo Data Science and Analytics Thrust The Hong Kong University of Science and Technology Guangzhou China Department of Computer Science Hong Kong Baptist University Hong Kong School of Computer Science and Technology Harbin Institute of Technology Shenzhen China Department of Computer Science and Engineering The Hong Kong University of Science and Technology Hong Kong

Sparsely-activated Mixture-of-Expert (MoE) layers have found practical applications in enlarging the model size of large-scale foundation models, with only a sub-linear increase in computation demands. Despite the wide adoption of hybrid parallel paradigms like model parallelism, expert parallelism, and expert-sharding parallelism (i.e., MP+EP+ESP) to support MoE model training on GPU clusters, the training efficiency is hindered by communication costs introduced by these parallel paradigms. To address this limitation, we propose Parm, a system that accelerates MP+EP+ESP training by designing two dedicated schedules for placing communication tasks. The proposed schedules eliminate redundant computations and communications and enable overlaps between intra-node and internode communications, ultimately reducing the overall training time. As the two schedules are not mutually exclusive, we provide comprehensive theoretical analyses and derive an automatic and accurate solution to determine which schedule should be applied in different scenarios. Experimental results on an 8-GPU server and a 32-GPU cluster demonstrate that Parm outperforms the state-of-the-art MoE training system, DeepSpeed-MoE, achieving 1.13× to 5.77× speedup on 1296 manually configured MoE layers and approximately 3× improvement on two real-world MoE models based on BERT and GPT-2. Copyright © 2024, The Authors. All rights reserved.

关键词： Machine learning

Shape-conditioned 3D Molecule Generation via Equivariant Diffusion Models

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Chen, Ziqi Peng, Bo Parthasarathy, Srinivasan Ning, Xia Computer Science and Engineering The Ohio Sate University ColumbusOH43210 United States Translational Data Analytics Institute The Ohio Sate University ColumbusOH43210 United States The Ohio Sate University ColumbusOH43210 United States

Ligand-based drug design aims to identify novel drug candidates of similar shapes with known active molecules. In this paper, we formulated an in silico shape-conditioned molecule generation problem to generate 3D molecule structures conditioned on the shape of a given molecule. To address this problem, we developed a translation- and rotation-equivariant shape-guided generative model ShapeMol. ShapeMol consists of an equivariant shape encoder that maps molecular surface shapes into latent embeddings, and an equivariant diffusion model that generates 3D molecules based on these embeddings. Experimental results show that ShapeMol can generate novel, diverse, drug-like molecules that retain 3D molecular shapes similar to the given shape condition. These results demonstrate the potential of ShapeMol in designing drug candidates of desired 3D shapes binding to protein target pockets. © 2023, CC BY-NC-SA.

关键词： Molecules

Point cloud registration for LiDAR and photogrammetric data: A critical synthesis and performance analysis on classic and deep learning algorithms

学校读者我要写书评

暂无评论

ISPRS Open Journal of Photogrammetry and Remote Sensing 2023年 8卷

作者： Xu, Ningli Qin, Rongjun Song, Shuang Geospatial Data Analytics Lab The Ohio State University Columbus United States Department of Civil Environmental and Geodetic Engineering The Ohio State University Columbus United States Department of Electrical and Computer Engineering The Ohio State University Columbus United States Translational Data Analytics Institute The Ohio State University Columbus United States

Three-dimensional (3D) point cloud registration is a fundamental step for many 3D modeling and mapping applications. Existing approaches are highly disparate in the data source, scene complexity, and application, therefore the current practices in various point cloud registration tasks are still ad-hoc processes. Recent advances in computer vision and deep learning have shown promising performance in estimating rigid/similarity transformation between unregistered point clouds of complex objects and scenes. However, their performances are mostly evaluated using a limited number of datasets from a single sensor (e.g. Kinect or RealSense cameras), lacking a comprehensive overview of their applicability in photogrammetric 3D mapping scenarios. In this work, we provide a comprehensive review of the state-of-the-art (SOTA) point cloud registration methods, where we analyze and evaluate these methods using a diverse set of point cloud data from indoor to satellite sources. The quantitative analysis allows for exploring the strengths, applicability, challenges, and future trends of these methods. In contrast to existing analysis works that introduce point cloud registration as a holistic process, our experimental analysis is based on its inherent two-step process to better comprehend these approaches including feature/keypoint-based initial coarse registration and dense fine registration through cloud-to-cloud (C2C) optimization. More than ten methods, including classic hand-crafted, deep-learning-based feature correspondence, and robust C2C methods were tested. We observed that the success rate of most of the algorithms are fewer than 40% over the datasets we tested and there are still are large margin of improvement upon existing algorithms concerning 3D sparse corresopondence search, and the ability to register point clouds with complex geometry and occlusions. With the evaluated statistics on three datasets, we conclude the best-performing methods for each step and prov

关键词： Deep learning Iterative closest point algorithm Point cloud registration Terrestrial laser scanning

M2: Mixed models with preferences, popularities and transitions for next-basket recommendation

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Peng, Bo Ren, Zhiyun Parthasarathy, Srinivasan Ning, Xia Department of Computer Science and Engineering The Ohio State University ColumbusOH43210 United States Department of Biomedical Informatics Translational Data Analytics Institute The Ohio State University ColumbusOH43210 United States Department of Biomedical Informatics The Ohio State University ColumbusOH43210 United States

Next-basket recommendation considers the problem of recommending a set of items into the next basket that users will purchase as a whole. In this paper, we develop a novel mixed model with preferences, popularities and transitions (M2) for the next-basket recommendation. This method models three important factors in next-basket generation process: 1) users' general preferences, 2) items' global popularities and 3) transition patterns among items. Unlike existing recurrent neural network-based approaches, M2does not use the complicated networks to model the transitions among items, or generate embeddings for users. Instead, it has a simple encoder-decoder based approach (ed-Trans) to better model the transition patterns among items. We compared M2with different combinations of the factors with 5 state-of-the-art next-basket recommendation methods on 4 public benchmark datasets in recommending the first, second and third next basket. Our experimental results demonstrate that M2significantly outperforms the state-of-the-art methods on all the datasets in all the tasks, with an improvement of up to 22.1%. In addition, our ablation study demonstrates that the ed-Trans is more effective than recurrent neural networks in terms of the recommendation performance. We also have a thorough discussion on various experimental protocols and evaluation metrics for the evaluation of next-basket recommendation methods. Copyright © 2020, The Authors. All rights reserved.

关键词： Recurrent neural networks

Completely self-supervised crowd counting via distribution matching

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Sam, Deepak Babu Agarwalla, Abhinav Joseph, Jimmy Sindagi, Vishwanath A. Babu, R. Venkatesh Patel, Vishal M. Video Analytics Lab Department of Computational and Data Sciences Indian Institute of Science Bangalore India Vision & Image Understanding Lab Department of Electrical and Computer Engineering Johns Hopkins University Baltimore United States

Dense crowd counting is a challenging task that demands millions of head annotations for training models. Though existing self-supervised approaches could learn good representations, they require some labeled data to map these features to the end task of density estimation. We mitigate this issue with the proposed paradigm of complete self-supervision, which does not need even a single labeled image. The only input required to train, apart from a large set of unlabeled crowd images, is the approximate upper limit of the crowd count for the given dataset. Our method dwells on the idea that natural crowds follow a power law distribution, which could be leveraged to yield error signals for backpropagation. A density regressor is first pretrained with self-supervision and then the distribution of predictions is matched to the prior by optimizing Sinkhorn distance between the two. Experiments show that this results in effective learning of crowd features and delivers significant counting performance. Furthermore, we establish the superiority of our method in less data setting as well. The code and models for our approach is available at https://***/val-iisc/css-ccnn. Copyright © 2020, The Authors. All rights reserved.

关键词： Deep learning