检索结果-内蒙古大学图书馆

Supervised Score-Based Modeling by Gradient Boosting

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Zhao, Changyuan Du, Hongyang Liu, Guangyuan Niyato, Dusit College of Computing and Data Science Nanyang Technological University Singapore CNRS@CREATE 1 Create Way # 08-01 Create Tower 138602 Singapore Department of Electrical and Electronic Engineering University of Hong Kong Hong Kong The Energy Research Institute @ NTU Interdisciplinary Graduate Program Singapore

Score-based generative models can effectively learn the distribution of data by estimating the gradient of the distribution. Due to the multi-step denoising characteristic, researchers have recently considered combining score-based generative models with the gradient boosting algorithm, a multi-step supervised learning algorithm, to solve supervised learning tasks. However, existing generative model algorithms are often limited by the stochastic nature of the models and the long inference time, impacting prediction performances. Therefore, we propose a Supervised Score-based Model (SSM), which can be viewed as a gradient boosting algorithm combining score matching. We provide a theoretical analysis of learning and sampling for SSM to balance inference time and prediction accuracy. Via the ablation experiment in selected examples, we demonstrate the outstanding performances of the proposed techniques. Additionally, we compare our model with other probabilistic models, including Natural Gradient Boosting (NGboost), Classification and Regression Diffusion Models (CARD), Diffusion Boosted Trees (DBT), and non-probabilistic GBM models. The experimental results show that our model outperforms existing models in both accuracy and inference time. © 2024, CC BY.

关键词： Self-supervised learning

Logarithmic-Regret Quantum Learning Algorithms for Zero-Sum Games

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Gao, Minbo Ji, Zhengfeng Li, Tongyang Wang, Qisheng State Key Laboratory of Computer Science Institute of Software Chinese Academy of Sciences Beijing China University of Chinese Academy of Sciences Beijing China Department of Computer Science and Technology Tsinghua University Beijing China Center on Frontiers of Computing Studies Peking University Beijing China School of Computer Science Peking University Beijing China Graduate School of Mathematics Nagoya University Nagoya Japan

We propose the first online quantum algorithm for solving zero-sum games with Oe(1) regret under the game setting.1 Moreover, our quantum algorithm computes an Ε-approximate Nash equilibrium of an m × n matrix zero-sum game in quantum time Oe(√m + n/Ε2.5). Our algorithm uses standard quantum inputs and generates classical outputs with succinct descriptions, facilitating end-to-end applications. Technically, our online quantum algorithm "quantizes" classical algorithms based on the optimistic multiplicative weight update method. At the heart of our algorithm is a fast quantum multi-sampling procedure for the Gibbs sampling problem, which may be of independent interest. Copyright © 2023, The Authors. All rights reserved.

关键词： Nash equilibrium

Deep Learning Approaches for Road Damage Detection Using YOLOv8 and Custom Convolutional Neural Network

学校读者我要写书评

暂无评论

Deep Learning Approaches for Road Damage Detection Using YOL...

Frontiers Technology of Information and computer (ICFTIC), IEEE International Conference on

作者： Muhammad Waseem Khan Khalid Mahmood Israr Hussain Muhammad Talha Qureshi Hafiz Muhammad Sanaullah Badar Wu Gao Center for Bio-inspired System-on-Chip School of Computer Science Northwestern Poly technical University Xi'an P. R. China Graduate School of Intelligent Data Science National Yunlin University of Science and Technology Taiwan R.O.C Institute of Computing MNS University of Agriculture Multan Pakistan Henan University Kaifeng China

ISBN: (数字)9798331541750

ISBN: (纸本)9798331541767

The identification of road damage is deemed very important to the preservation of infrastructure. Since recent development in deep learning provide great potential approaches, in this paper, three deep learning models, namely EfficientNet, CNN and YOLOv8 are considered and their performance is evaluated on a dataset with 8,586 images with labelled road damages consisting of seven subclasses including longitudinal cracks and potholes. Besides, other issues like class imbalance and variabilities in the environment were studied and considered in its assessment based on the Models accuracy, precision, recall, and F1-score. With an accuracy of 64%, YOLOv8 is the most accurate model for identifying various types of damages, particularly in identifying multiple damages present in one or more types of damage, as was seen in the high accuracy rate of a split sample of 44 images of multiple damages. Our findings showcase YOLOv8 as a strong contender for real-time road damage detection with potential of enhancement with better augmentation and fine-tuning.

关键词： Deep learning Location awareness Accuracy Roads Image edge detection Transfer learning Real-time systems Data models Convolutional neural networks Reliability

A Hydrodynamic Analysis of Water System in Dadahup Swamp Irrigation Area 1

学校读者我要写书评

暂无评论

A Hydrodynamic Analysis of Water System in Dadahup Swamp Irr...

1st International Conference on computer science and Artificial Intelligence, ICCSAI 2021

作者： Purboseno, Sentot Suparyanto, Teddy Hidayat, Alam Ahmad Pardamean, Bens Institute of Agriculture STIPER Department of Agricultural Engineering Yogyakarta55283 Indonesia Bina Nusantara University Bioinformatics and Data Science Research Center Jakarta11480 Indonesia Bina Nusantara University Bioinformatics and Data Science Research Center BINUS Graduate Program-Master of Computer Science Computer Science Department Jakarta11480 Indonesia

ISBN: (纸本)9781728133331

Dadahup Swamp Irrigation Area (DIR) in Kapuas Regency, Central Kalimantan is developed for agricultural activities to provide food security after the pandemic. The water system consists of various channels, gates, and a flood embankment structure, which is also connected to three different rivers: Barito River, Mangkatif River, and Kapuas Mumpung River. To quantify flood risk assessments in the area, we perform a hydrodynamic analysis using the unsteady flow method via HEC-RAS software version 4.1.0. We use geometrical profiles from topography measurements and the outer boundary of the Dadahup DIR as input for the hydrologic model. Our results estimate the maximum surface water level from both cross and longitudinal sections in the channels and find that several points near the flood embankment area are prone to flood events. Our modeling approach provides a preliminary assessment for the local government to formulate a flood mitigation plan and policy in the area. © 2021 IEEE.

关键词： Rivers

Towards Adversarial Robustness for Multi-Mode Data through Metric Learning

学校读者我要写书评

暂无评论

Sensors 2023年第13期23卷 6173页

作者： Khan, Sarwar Chen, Jun-Cheng Liao, Wen-Hung Chen, Chu-Song Research Center for Information Technology Innovation Academia Sinica Taipei11529 Taiwan Social Networks Human-Centered Computing Taiwan International Graduate Program Academia Sinica Taipei11529 Taiwan Department of Computer Science National Chengchi University Taipei11605 Taiwan Department of Computer Science and Information Engineering National Taiwan University Taipei106319 Taiwan

Adversarial attacks have become one of the most serious security issues in widely used deep neural networks. Even though real-world datasets usually have large intra-variations or multiple modes, most adversarial defense methods, such as adversarial training, which is currently one of the most effective defense methods, mainly focus on the single-mode setting and thus fail to capture the full data representation to defend against adversarial attacks. To confront this challenge, we propose a novel multi-prototype metric learning regularization for adversarial training which can effectively enhance the defense capability of adversarial training by preventing the latent representation of the adversarial example changing a lot from its clean one. With extensive experiments on CIFAR10, CIFAR100, MNIST, and Tiny ImageNet, the evaluation results show the proposed method improves the performance of different state-of-the-art adversarial training methods without additional computational cost. Furthermore, besides Tiny ImageNet, in the multi-prototype CIFAR10 and CIFAR100 where we reorganize the whole datasets of CIFAR10 and CIFAR100 into two and ten classes, respectively, the proposed method outperforms the state-of-the-art approach by 2.22% and 1.65%, respectively. Furthermore, the proposed multi-prototype method also outperforms its single-prototype version and other commonly used deep metric learning approaches as regularization for adversarial training and thus further demonstrates its effectiveness. © 2023 by the authors.

关键词： Deep neural networks

HAAQI-Net: A Non-intrusive Neural Music Audio Quality Assessment Model for Hearing Aids

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Wisnu, Dyah A.M.G. Rini, Stefano Zezario, Ryandhimas E. Wang, Hsin-Min Tsao, Yu Taiwan International Graduate Program—Social Network and Human Centered Computing Institute of Information Science Academia Sinica Taipei Taiwan College of Informatics National Chengchi University Taipei Taiwan Institute of Communications Engineering National Yang Ming Chiao Tung University Hsinchu Taiwan Research Center for Information Technology Innovation Academia Sinica Taipei Taiwan Institute of Information Science Academia Sinica Taipei Taiwan

This paper introduces HAAQI-Net, a non-intrusive deep learning-based music audio quality assessment model for hearing aid users. Unlike traditional methods like the Hearing Aid Audio Quality Index (HAAQI) that require intrusive reference signal comparisons, HAAQI-Net offers a more accessible and computationally efficient alternative. By utilizing a Bidirectional Long Short-Term Memory (BLSTM) architecture with attention mechanisms and features extracted from the pre-trained BEATs model, it can predict HAAQI scores directly from music audio clips and hearing loss patterns. Experimental results demonstrate HAAQI-Net’s effectiveness, achieving a Linear Correlation Coefficient (LCC) of 0.9368, a Spearman’s Rank Correlation Coefficient (SRCC) of 0.9486, and a Mean Squared Error (MSE) of 0.0064 and inference time significantly reduces from 62.52 to 2.54 seconds. To address computational overhead, a knowledge distillation strategy was applied, reducing parameters by 75.85% and inference time by 96.46%, while maintaining strong performance (LCC: 0.9071, SRCC: 0.9307, MSE: 0.0091). To expand its capabilities, HAAQI-Net was adapted to predict subjective human scores like the Mean Opinion Score (MOS) through fine-tuning. This adaptation significantly improved prediction accuracy, validated through statistical analysis. Furthermore, the robustness of HAAQI-Net was evaluated under varying Sound Pressure Level (SPL) conditions, revealing optimal performance at a reference SPL of 65 dB, with accuracy gradually decreasing as SPL deviated from this point. The advancements in subjective score prediction, SPL robustness, and computational efficiency position HAAQI-Net as a scalable solution for music audio quality assessment in hearing aid applications, contributing to efficient and accurate models in audio signal processing and hearing aid technology. Copyright © 2024, The Authors. All rights reserved.

关键词： Mean square error

Improving Land Cover Segmentation Using Multispectral Dataset

学校读者我要写书评

暂无评论

Improving Land Cover Segmentation Using Multispectral Datase...

International Conference on Informatics and computing (ICIC)

作者： Herlawati Herlawati Rahmadya Trias Handayanto Yaya Heryadi Edi Abdurachman Haryono Soeparno Edy Irwansyah Eka Miranda Informatics Universitas Bhayangkara Jakarta Raya Jakarta Indonesia Computer Engineering Universitas Islam 45 Bekasi Indonesia Computer Science Department BINUS Graduate Program – Doctor of Computer Science Bina Nusantara University Jakarta Indonesia Trisakti Institute of Transportation and Logistic Jakarta Indonesia Computer Science Department School of Computer Science Bina Nusantara University Jakarta Indonesia Information Systems Department School of Information Systems Bina Nusantara University Jakarta Indonesia

computer vision has been used in many areas such as medical, transportation, military, geography, etc. The fast development of sensor devices inside camera and satellite provides not only red-greed-blue (RGB) images but also multispectral dataset with some channels including RGB, infrared, short-wave, and thermal wave. Most of the dataset is panchromatic (black and white) and RGB, for example Google Map and other satellite-based map applications. This study examines the effects of multispectral dataset for semantic segmentation of land cover. The comparison between RGB with band 2 to band 7 of Landsat 8 Satellite shows an improvement of accuracy from 90.283 to 94.473 for U-Net and from 91.76 to 95.183 for DeepLabV3+. In addition, this research also compares two well-known semantic segmentation methods, namely U-Net and DeepLabV3+, that shown that DeepLabV3+ outperformed U-Net regarding to speed and accuracy. Testing was conducted in the Karawang Regency area, West Java, Indonesia.

关键词：

Deep and shallow data science for multi-scale optical neuroscience

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Mishne, Gal Charles, Adam Halıcıoğlu Data Science Institute Department of Electrical and Computer Engineering the Neurosciences Graduate Program UC San Diego 9500 Gilman Drive La Jolla CA92093 United States Department of Biomedical Engineering Kavli Neuroscience Discovery Institute Center for Imaging Science Department of Neuroscience Mathematical Institute for Data Science Johns Hopkins University BaltimoreMD21287 United States

Optical imaging of the brain has expanded dramatically in the past two decades. New optics, indicators, and experimental paradigms are now enabling in-vivo imaging from the synaptic to the cortex-wide scales. To match the resulting flood of data across scales, computational methods are continuously being developed to meet the need of extracting biologically relevant information. In this pursuit, challenges arise in some domains (e.g., SNR and resolution limits in micron-scale data) that require specialized algorithms. These algorithms can, for example, make use of state-of-the-art machine learning to maximally learn the details of a given scale to optimize the processing pipeline. In contrast, other methods, however, such as graph signal processing, seek to abstract away from some of the details that are scale-specific to provide solutions to specific sub-problems common across scales of neuroimaging. Here we discuss limitations and tradeoffs in algorithmic design with the goal of identifying how data quality and variability can hamper algorithm use and dissemination. © 2024, CC BY.

关键词： Neuroimaging

A Robust Key Exchange and Tamper-Resistant Protocol for HAN and NAN Networks in Smart Grids

学校读者我要写书评

暂无评论

IEEE Internet of Things Journal 2025年第12期12卷 21394-21403页

作者： Ayub, Muhammad Faizan Li, Xiong Mahmood, Khalid Shamshad, Salman Das, Ashok Kumar Wang, Guijuan University of Electronic Science and Technology of China School of Computer Science and Engineering Sichuan Chengdu611731 China National Yunlin University of Science and Technology Graduate School of Intelligent Data Science Yunlin Douliu64002 Taiwan The University of Lahore Department of Software Engineering Lahore54590 Pakistan International Institute of Information Technology Center for Security Theory and Algorithmic Research Hyderabad500 032 India Korea University College of Informatics Department of Computer Science and Engineering Anam-ro Seoul145 Korea Republic of Qilu University of Technology Shandong Academy of Sciences Key Laboratory of Computing Power Network and Information Security Ministry of Education Shandong Computer Science Center Jinan250353 China Shandong Fundamental Research Center for Computer Science Shandong Provincial Key Laboratory of Industrial Network and Information System Security Jinan250353 China

Smart Grids (SG) rely on Home Area Networks (HAN) and Neighborhood Area Networks (NAN) to ensure efficient power distribution, real-time monitoring, and seamless communication between smart devices. Despite these advantages, the use of public communication channels in HAN and NAN networks introduces critical challenges, such as vulnerability to impersonation, physical tampering, and scalability issues in resource-constrained environments. These issues compromise the stability, reliability, and security of SG environments. Existing protocols often fail to adequately address these challenges, particularly in ensuring resistance to physical tampering of supply lines and impersonation attacks. Additionally, they struggle to minimize the computation, communication, and energy costs associated with securing a resource-constrained SG environment. Therefore, we propose a robust key exchange and tamper-resistant protocol for HAN and NAN networks to address these limitations. The proposed protocol leverages the Physical Unclonable Function (PUF) mechanism to offer physical tampering resistance to smart meters. We validate our protocol formally using the Random or Real (RoR) model, which confirms its security robustness. Additionally, our informal security analysis highlights the protocol's resilience against impersonation, physical tampering attacks, etc. We analyze and compare the performance of the proposed protocol, which further demonstrates our protocol's effectiveness and security compared to competing protocols. Moreover, the proposed protocol achieves resource efficiency with 45.99% and 58.85% substantial reductions in computation overhead and energy overhead, respectively, compared to competing protocols. These results showcase the protocol's enhanced security and practicality for resource-constrained SG environments. © 2014 IEEE.

关键词： Authentication Protocol