检索结果-内蒙古大学图书馆

arXiv 2023年

作者： Takahashi, Kazuki Fukai, Tomoki Sakai, Yutaka Takekawa, Takashi Informatics Program Graduate School of Engineering Kogakuin University of Technology and Engineering Japan Neural Coding and Brain Computing Unit Okinawa Institute of Science and Technology Japan Brain Science Institute Tamagawa University Japan

The agent learns to organize decision behavior to achieve a behavioral goal, such as reward maximization, and reinforcement learning is often used for this optimization. Learning an optimal behavioral strategy is difficult under the uncertainty that events necessary for learning are only partially observable, called as Partially Observable Markov Decision Process (POMDP). However, the real-world environment also gives many events irrelevant to reward delivery and an optimal behavioral strategy. The conventional methods in POMDP, which attempt to infer transition rules among the entire observations, including irrelevant states, are ineffective in such an environment. Supposing Redundantly Observable Markov Decision Process (ROMDP), here we propose a method for goal-oriented reinforcement learning to efficiently learn state transition rules among reward-related"core states" from redundant observations. Starting with a small number of initial core states, our model gradually adds new core states to the transition diagram until it achieves an optimal behavioral strategy consistent with the Bellman equation. We demonstrate that the resultant inference model outperforms the conventional method for POMDP. We emphasize that our model only containing the core states has high explainability. Furthermore, the proposed method suits online learning as it suppresses memory consumption and improves learning speed. Copyright © 2023, The Authors. All rights reserved.

关键词： Reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

A survey on deep geometry learning:From a representation perspective

引用

Computational Visual Media 2020年第2期6卷 113-133页

作者： Yun-Peng Xiao Yu-Kun Lai Fang-Lue Zhang Chunpeng Li Lin Gao Institute of Computing Technology Chinese Academy of SciencesBeijingChina School of Computer Science and Informatics Cardiff UniversityWalesUK School of Engineering and Computer Science Victoria University of WellingtonWellingtonNew Zealand

Researchers have achieved great success in dealing with 2 D images using deep *** recent years,3 D computer vision and geometry deep learning have gained ever more *** advanced techniques for 3 D shapes have been proposed for different *** 2 D images,which can be uniformly represented by a regular grid of pixels,3 D shapes have various representations,such as depth images,multi-view images,voxels,point clouds,meshes,implicit surfaces,*** performance achieved in different applications largely depends on the representation used,and there is no unique representation that works well for all ***,in this survey,we review recent developments in deep learning for 3 D geometry from a representation perspective,summarizing the advantages and disadvantages of different representations for different *** also present existing datasets in these representations and further discuss future research directions.

关键词： 3D shape representation geometry learning neural networks computer graphics

来源：评论

学校读者我要写书评

暂无评论

Evaluation of Atmospheric Pollution and Estimation of Remaining Atmospheric Environmental Capacity in Xuzhou City

Evaluation of Atmospheric Pollution and Estimation of Remain...

引用

IEEE International Symposium on Geoscience and Remote Sensing (IGARSS)

作者： Shuhui Wu Yong Xue Yuxin Sun Chunlin Jin Xingxing Jiang Xi Lu Wenhao Liu School of Environment and Spatial Informatics China University of Mining and Technology Xuzhou China Artificial Intelligence Research Institute China University of Mining and Technology Xuzhou China School of Computing and Mathematics College of Science and Engineering University of Derby Derby UK

Based on the data of air quality monitoring stations and high-resolution remote sensing product data in Xuzhou City, the atmospheric pollutants in 2020 and 2021 were analyzed. The modified A-values method and the model simulation method were used to estimate the remaining atmospheric environmental capacity (RAEC) of PM 10 and PM 2.5 pollutants in Xuzhou, Jiangsu Province. The results show that the excessive PM 10 and PM 2.5 pollutants are the main problems of atmospheric pollution in Xuzhou, and the annual emissions still need to be reduced are 10.80×10 4 t/a and 5.98×10 4 t/a, respectively. Among them, the situation is the most serious in Tongshan District, which still needs to cut annual PM 10 and PM 2.5 emissions by 7.41×10 4 t/a and 3.70×10 4 t/a, respectively. Xinyi City has the smallest annual PM 10 emission reduction, which is 0.89×10 4 t/a. Suining County needs to cut the smallest PM 2.5 emissions, at 0.67×10 4 t/a. In addition, there are significant quarterly differences in RAEC, with the first quarter > the fourth quarter > the second quarter > the third quarter. Except for the third quarter, the excess atmospheric environmental capacity was the most serious in Xuzhou urban area, the other three quarters were the highest in Tongshan District.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Support Vector Machine for Satellite Images Classification Using Radial Basis Function Kernel Method 9th

Support Vector Machine for Satellite Images Classification U...

引用

9th International Conference on computing and informatics, ICOCI 2023

作者： Mansor, Nur Suhaili Awang, Hapini Malami, Sarkin Tudu Shehu Zolkafli, Amirulikhsan Taiye, Mohammed Ahmed Maulana, Hanhan Institute for Advanced and Smart Digital Opportunities School of Computing Universiti Utara Malaysia Kedah Sintok06010 Malaysia Department of Computer Science and Information Technology Faculty of Science Sokoto State University Sokoto Sokoto852101 Nigeria Department of Cultural Sciences Faculty of Arts and Humanities Linnaeus University Vaxjo351 95 Sweden Department of Informatics Engineering and Computer Science Universitas Komputer Indonesia Bandung40134 Indonesia

ISBN: (纸本)9789819995882

Machine learning, particularly Support Vector Machines (SVM), has gained popularity in geospatial data processing and image classification. Geospatial data from various sources may contain errors, impacting image classification accuracy. Traditional pixel-based and object-based methods struggle to classify complex land cover classes accurately. Previous studies explored machine learning algorithms like Random Forests, K-Nearest Neighbors, and Neural Networks. Still, they faced challenges capturing intricate relationships within images and required substantial labeled training data, leading to computational expenses. SVM with polynomial kernels was attempted in some studies, but it suffered from potential overfitting and inefficiency for large datasets. To overcome these issues, this study employed SVM with RBF and Linear kernels to classify multispectral satellite images from the SPOT-6 Satellite Imagery dataset in Sungai Kelang, Malaysia. Previous research evaluated each kernel’s performance accuracy compared using a test dataset, utilizing open-source tools like Jupyter Notebooks and Python libraries to explore SVM’s potential as a high-performance satellite image classification technique. The findings revealed that SVM with RBF kernel outperformed SVM with polynomial or linear kernels in classifying satellite images. The RBF kernel’s robustness allowed SVM to model intricate decision boundaries and capture complex patterns in the image data, making it suitable for tasks with non-linearly separable data. The study introduces a new methodology and theoretical contribution to image classification-related literature, shedding light on the efficacy of SVM-RBF for geospatial data processing. It provides an alternative to traditional approaches for complex image classification tasks. Moreover, the research assists in selecting the optimal algorithm for remote sensing and satellite imagery applications. © 2024, The Author(s), under exclusive license to Springer Nature Si

关键词： Image classification

来源：评论

学校读者我要写书评

暂无评论

Optimized circuits for windowed modular arithmetic with applications to quantum attacks against RSA

arXiv

引用

arXiv 2025年

作者： Luongo, Alessandro Narasimhachar, Varun Sireesh, Adithya Centre for Quantum Technologies National University of Singapore Singapore Inveriant Pte. Ltd. Singapore Institute of High Performance Computing Agency for Science Technology and Research Singapore School of Informatics University of Edinburgh Scotland United Kingdom

Windowed arithmetic [Gidney, 2019] is a technique for reducing the cost of quantum arithmetic circuits with space–time tradeoffs using memory queries to precomputed tables. It can reduce the asymptotic cost of modular exponentiation from O(n3) to O(n3/log2 n) operations, resulting in the current state-of-the-art compilations of quantum attacks against modern cryptography. In this work we introduce four optimizations to windowed modular exponentiation. We (1) show how the cost of unlookups can be reduced by 66% asymptotically in the number of bits, (2) illustrate how certain addresses can be bypassed, reducing both circuit depth and the overall lookup cost, (3) demonstrate that multiple lookup–addition operations can be merged into a single, larger lookup at the start of the modular exponentiation circuit, and (4) reduce the depth of the unary conversion for unlookups. On a logical level, this leads to a 3% improvement in Toffoli count and Toffoli depth for modular exponentiation circuits relevant to cryptographic applications. This translates to some improvements on [Gidney and Ekerå, 2021]’s factoring algorithm: for a given number of physical qubits, our improvements show a reduction in the expected runtime from 2% to 6% for factoring RSA-2048 *** Codes 81P68, 68Q09, 68Q12 © 2025, CC0.

关键词： Table lookup

来源：评论

学校读者我要写书评

暂无评论

Research on the Application of LSTM-SA-AdaBoost Hybrid Model in Stock Forecasting

Research on the Application of LSTM-SA-AdaBoost Hybrid Model...

引用

International Conference on Applied Machine Learning (ICAML)

作者： Yu Sun Sofianita Mutalib Nasiroh Omar Mingjing Huang Guangdong University of Science and Technology Dongguan Guangdong Province China School of Computing Sciences College of Computing Informatics and Media Universiti Teknologi MARA Shah Alam Selangor Malaysia School of Electrical Engineering College of Engineering Universiti Teknologi MARA Shah Alam Selangor Malaysia

To improve the fitting and accuracy of stock prediction, an improved deep neural network combined with AdaBoost model (LSTM-SA-AdaBoost) is proposed. The model feature engineering includes data cleaning, correlation analysis and normalization. The model uses simulated annealing algorithm to optimize the model parameters. The attributes after feature selection will be trained, and the predicted attributes will be predicted and optimized iteratively through the two-layer LSTM network. From the experiment results, it shows that LSTM-SA-AdaBoost algorithm is superior to the unmodified LSTM-AdaBoost model and LSTM-XGBoost model, compared with the single-target feature selection algorithm of LSTM and RNN network models, it has better fitting and better accuracy.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Mind the Gap: Confidence Discrepancy Can Guide Federated Semi-Supervised Learning Across Pseudo-Mismatch

arXiv

引用

arXiv 2025年

作者： Liu, Yijie Shang, Xinyi Zhang, Yiqun Lu, Yang Gong, Chen Xue, Jing-Hao Wang, Hanzi Key Laboratory of Multimedia Trusted Perception and Efficient Computing Ministry of Education of China Xiamen University Xiamen China Fujian Key Laboratory of Sensing and Computing for Smart City School of Informatics Xiamen University Xiamen China Department of Statistical Science University College London United Kingdom School of Computer Science and Technology Guangdong University of Technology Guangzhou China Department of Automation Shanghai Jiao Tong University China

Federated Semi-Supervised Learning (FSSL) aims to leverage unlabeled data across clients with limited labeled data to train a global model with strong generalization ability. Most FSSL methods rely on consistency regularization with pseudo-labels, converting predictions from local or global models into hard pseudo-labels as supervisory signals. However, we discover that the quality of pseudo-label is largely deteriorated by data heterogeneity, an intrinsic facet of federated learning. In this paper, we study the problem of FSSL in-depth and show that (1) heterogeneity exacerbates pseudo-label mismatches, further degrading model performance and convergence, and (2) local and global models’ predictive tendencies diverge as heterogeneity increases. Motivated by these findings, we propose a simple and effective method called Semi-supervised Aggregation for Globally-Enhanced Ensemble (SAGE), that can flexibly correct pseudo-labels based on confidence discrepancies. This strategy effectively mitigates performance degradation caused by incorrect pseudo-labels and enhances consensus between local and global models. Experimental results demonstrate that SAGE outperforms existing FSSL methods in both performance and convergence. Our code is available at https://***/Jay-Codeman/SAGE. Copyright © 2025, The Authors. All rights reserved.

关键词： Federated learning

来源：评论

学校读者我要写书评

暂无评论

StyleGAN-∞: Extending StyleGAN to Arbitrary-Ratio Translation with StyleBook

引用

IEEE Transactions on Visualization and Computer Graphics 2024年

作者： Dai, Yihua Xiang, Tianyi Deng, Bailin Du, Yong Cai, Hongmin Qin, Jing He, Shengfeng South China University Of Technology School Of Computer Science And Engineering Guangzhou China Singapore Management University School Of Computing And Information Systems Singapore Cardiff University School Of Computer Science And Informatics United Kingdom Ocean University Of China Faculty Of Information Science And Engineering Qingdao China Hong Kong Polytechnic University School Of Nursing Hong Kong Hong Kong

Although pre-trained large-scale generative models StyleGAN series have proven to be effective in various editing and translation tasks, they are limited to pre-defined fixed aspect ratio. To overcome this limitation, we propose StyleGAN-∞, a model that enables pre-trained StyleGAN to perform arbitrary-ratio conditional synthesis. Our key insight is to distill the expressive StyleGAN features into a StyleBook, such that an arbitrary-ratio condition can be translated to other forms by properly assembling pre-defined StyleBook vectors. To learn and leverage the StyleBook, we employ a network with three distinct stages, each corresponding to StyleBook extraction, StyleBook correspondence learning, and arbitrary-ratio synthesis. Extensive experiments on various conditional synthesis tasks, like super-resolution, sketch synthesis, and semantic synthesis, demonstrate superior performances over state-of-the-art image-to-image translation methods. Moreover, our model can easily generate megapixel images in diverse modalities by taking advantage of different pre-trained StyleGAN models. © 1995-2012 IEEE.

关键词： Generative adversarial networks

来源：评论

学校读者我要写书评

暂无评论

Factors influencing SaaS adoption by MSMEs 7

Factors influencing SaaS adoption by MSMEs

引用

7th International Conference on Software Engineering and Computer Systems and 4th International Conference on Computational Science and Information Management, ICSECS-ICOCSIM 2021

作者： Ayuningtyas, Ayuningtyas Azmi, Zafril Rizal M Abdullah Arshah, Ruzaini Ahmad, Muhammad Aliif Khubrani, Mousa Faculty of Technology and Informatics Universitas Dinamika Surabaya East Java Indonesia Faculty of Computing Universiti Malaysia Pahang Malaysia School of Computing Faculty of Engineering Universiti Teknologi Malaysia Malaysia Faculty of Computer Science Information Technology Jazan University Saudi Arabia

ISBN: (纸本)9781665414074

Cloud computing has been around for quite long, yet some organizations have never used it, especially MSMEs in Indonesia. Several factors hinder its implementation by MSMEs, including unequal infrastructure for using the internet in Indonesia. This paper assesses the literature on cloud computing, SaaS, e-Commerce, and e-wallet to determine the initial factors that affect technology implementation. The results showed that the frameworks widely used to obtain and assess variables include the DOI, TAM, and TOE. Furthermore, compatibility and the pressure from competitors to adopt technology are the highly considered variables. Among the factors, perceived usefulness and ease of use is also deemed by MSMEs in implementing e-commerce and e-wallet. © 2021 IEEE.

关键词： Software as a service (SaaS)

来源：评论

学校读者我要写书评

暂无评论

FLGO: A Fully Customizable Federated Learning Platform

arXiv

引用

arXiv 2023年

作者： Wang, Zheng Fan, Xiaoliang Peng, Zhaopeng Li, Xueheng Yang, Ziqi Feng, Mingkuan Yang, Zhicheng Liu, Xiao Wang, Cheng Fujian Key Laboratory of Sensing and Computing for Smart Cities School of Informatics Xiamen University Xiamen China Key Laboratory of Multimedia Trusted Perception and Efficient Computing Ministry of Education of China Xiamen University Xiamen China School of Information Technology Deakin University Geelong Australia

Federated learning (FL) has found numerous applications in healthcare, finance, and IoT scenarios. Many existing FL frameworks offer a range of benchmarks to evaluate the performance of FL under realistic conditions. However, the process of customizing simulations to accommodate application-specific settings, data heterogeneity, and system heterogeneity typically remains unnecessarily complicated. This creates significant hurdles for traditional ML researchers in exploring the usage of FL, while also compromising the shareability of codes across FL frameworks. To address this issue, we propose a novel lightweight FL platform called FLGo, to facilitate cross-application FL studies with a high degree of shareability. Our platform offers 40+ benchmarks, 20+ algorithms, and 2 system simulators as out-of-the-box plugins. We also provide user-friendly APIs for quickly customizing new plugins that can be readily shared and reused for improved reproducibility. Finally, we develop a range of experimental tools, including parallel acceleration, experiment tracker and analyzer, and parameters auto-tuning. FLGo is maintained at ***. © 2023, CC BY.

关键词： Benchmarking

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：