检索结果-内蒙古大学图书馆

SSRN 2022年

作者： Fang, Aiqing Wu, Junsheng Ying, Li School of Computer Science and Engineering Northwestern Polytechnical University Xi’an710072 China School of Software Northwestern Polytechnical University Xi’an710072 China

Deep learning-based methods generally adopt an unsupervised similarity fusion optimization mechanism (UFM) or Generative Adversarial Network-based fusion optimization mechanism (GFM) toward image fusion tasks without ground truth (GT) labels, which leads to the problem that the fused image is over smooth or under smooth. This paper adopts a simple-yet-effective self-supervised fusion representation learning optimization mechanism (FLM) by directly maximizing the mutual information between the fused image and the samples (i.e., positive and negative samples), which aims to overcome over- and under-smoothing problems at intra-image level. Following InfoNCE, we further propose a smooth representation learning optimization mechanism (SFLM), which balances the fused image under-smoothing and over-smoothing at the inter-image level. Specifically, the SFLM is optimized by minimizing the total variation loss in the pixel space and maximizing the total variation loss in the gradient space. To show the superiority of SFLM, quantitative and qualitative experiments are carried out on four public infrared and visible image fusion datasets, including RoadScene, TNO, VIFB, and MF. Experimental results and comparisons with the state-of-the-art on four public datasets validate the effectiveness, robustness, and overall superiority of our method. © 2022, The Authors. All rights reserved.

关键词： Image fusion

来源：评论

学校读者我要写书评

暂无评论

Securely Fine-tuning Pre-trained Encoders Against Adversarial Examples

arXiv

引用

arXiv 2024年

作者： Zhou, Ziqi Li, Minghui Liu, Wei Hu, Shengshan Zhang, Yechao Wan, Wei Xue, Lulu Zhang, Leo Yu Yao, Dezhong Jin, Hai National Engineering Research Center for Big Data Technology and System China Services Computing Technology and System Lab. China Cluster and Grid Computing Lab. China Hubei Engineering Research Center on Big Data Security China Hubei Key Laboratory of Distributed System Security China School of Computer Science and Technology Huazhong University of Science and Technology China School of Software Engineering Huazhong University of Science and Technology China School of Cyber Science and Engineering Huazhong University of Science and Technology China School of Information and Communication Technology Griffith University Australia

With the evolution of self-supervised learning, the pre-training paradigm has emerged as a predominant solution within the deep learning landscape. Model providers furnish pre-trained encoders designed to function as versatile feature extractors, enabling downstream users to harness the benefits of expansive models with minimal effort through fine-tuning. Nevertheless, recent works have exposed a vulnerability in pre-trained encoders, highlighting their susceptibility to downstream-agnostic adversarial examples (DAEs) meticulously crafted by attackers. The lingering question pertains to the feasibility of fortifying the robustness of downstream models against DAEs, particularly in scenarios where the pre-trained encoders are publicly accessible to the attackers. In this paper, we initially delve into existing defensive mechanisms against adversarial examples within the pre-training paradigm. Our findings reveal that the failure of current defenses stems from the domain shift between pre-training data and downstream tasks, as well as the sensitivity of encoder parameters. In response to these challenges, we propose Genetic Evolution-Nurtured Adversarial Fine-tuning (Gen-AF), a two-stage adversarial fine-tuning approach aimed at enhancing the robustness of downstream models. Gen-AF employs a genetic-directed dual-track adversarial fine-tuning strategy in its first stage to effectively inherit the pre-trained encoder. This involves optimizing the pre-trained encoder and classifier separately while incorporating genetic regularization to preserve the model’s topology. In the second stage, Gen-AF assesses the robust sensitivity of each layer and creates a dictionary, based on which the top-k robust redundant layers are selected with the remaining layers held fixed. Upon this foundation, we conduct evolutionary adaptability fine-tuning to further enhance the model’s generalizability. Our extensive experiments, conducted across ten self-supervised training methods and six

关键词： Topology

来源：评论

学校读者我要写书评

暂无评论

E2S2: Encoding-Enhanced Sequence-to-Sequence Pretraining for Language Understanding and Generation

arXiv

引用

arXiv 2022年

作者： Zhong, Qihuang Ding, Liang Liu, Juhua Du, Bo Tao, Dacheng The School of Computer Science National Engineering Research Center for Multimedia Software Institute of Artificial Intelligence Hubei Key Laboratory of Multimedia and Network Communication Engineering Wuhan University Wuhan China The School of Computer Science Faculty of Engineering The University of Sydney Australia

Sequence-to-sequence (seq2seq) learning is a popular fashion for large-scale pretraining language models. However, the previous seq2seq pretraining models generally focus on reconstructive objectives on the decoder side and neglect the effect of encoder-side supervision, which we argue may lead to sub-optimal performance. To verify our hypothesis, we first empirically study the functionalities of the encoder and decoder in seq2seq pretrained language models, and find that the encoder takes an important but under-exploitation role than the decoder regarding the downstream performance and neuron activation. Therefore, we propose an encoding-enhanced seq2seq pretraining strategy, namely E2S2, which improves the seq2seq models via integrating more efficient self-supervised information into the encoders. Specifically, E2S2 adopts two self-supervised objectives on the encoder side from two aspects: 1) locally denoising the corrupted sentence (denoising objective);and 2) globally learning better sentence representations (contrastive objective). With the help of both objectives, the encoder can effectively distinguish the noise tokens and capture high-level (i.e., syntactic and semantic) knowledge, thus strengthening the ability of seq2seq model to accurately achieve the conditional generation. On a large diversity of downstream natural language understanding and generation tasks, E2S2 dominantly improves the performance of its powerful backbone models, e.g., BART and T5. For example, upon BART backbone, we achieve +1.1% averaged gain on the general language understanding evaluation (GLUE) benchmark and +1.75% F0.5 score improvement on CoNLL2014 dataset. We also provide in-depth analyses to show the improvement stems from better linguistic representation. We hope that our work will foster future self-supervision research on seq2seq language model pretraining. Copyright © 2022, The Authors. All rights reserved.

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

The functionality on netpad 16

The functionality on netpad

引用

16th IEEE International Conference on computer science and Education, ICCSE 2021

作者： Li, Ning Li, Chuangzhong Gao, Ying Wang, Ying Zhao, Jietong Zhang, Linjing Guangzhou University School of Computer Science and Cyber Engineering Guangzhou China Institute of Guangzhou Software Engineering Guangzhou China Institute of Computing Science and Technology Guangzhou University Guangzhou China Fuzhou Administration Institute Fuzhou China

ISBN: (纸本)9781665414685

Modern information technology-assisted teaching is an important teaching means. In mathematics teaching activities, graphical changes can be vividly displayed through dynamic geometry software (DGS). Function is the basic knowledge in mathematics, and the functionality is the basic component of DGS. NetPad, one of the DGS, has more performance and simpler operation compared with other DGS. In this paper, we introduce the functionality of NetPad. It includes user-defined coordinate system, prompt function, expression input, implicit function, polar equation, parameter equation, function surface etc.. And its operation procedure is simpler to drawing curves dynamic trajectory and exploring function curves intersection. Through a teaching case, the superiority of NetPad in functionality is verified. © 2021 IEEE.

关键词： Functions

来源：评论

学校读者我要写书评

暂无评论

An Unsupervised Convolutional Adversarial Anomaly Detection Model for IoT Data Infrastructure

An Unsupervised Convolutional Adversarial Anomaly Detection ...

引用

IEEE International Conference on High Performance Computing and Communications (HPCC)

作者： Peian Wen Zhenyu Yang Lei Wu Sibo Qi Juan Chen School of Computer and Software Engineering Xihua University Chengdu China School of Mathematical Sciences University of Electronic Science and Technology of China Chengdu China

Anomaly detection is critical to ensure the IoT (Internet of Things) data infrastructures' Quality of Service. However, due to the complexity of incon-spicuous(indistinct) anomalies, high dynamicity, and lack of anomaly labels in the operational IoT systems and cloud infrastructures, multivariate time series anomaly detection becomes more difficult. Existing approaches are rarely effective in meeting these challenges. In this paper, We propose a novel convolutional adversarial model based on CNN adversarial training with POT (Peak Over Threshold) dynamic threshold selection (CAT-POT) for multivariate time series anomaly detection in the IoT data. We build on the original autoencoder, replacing the fundamental linear changes with a convolutional neural network and adding a convolutional decoder to make an adversarial training architecture. The ability of the model to identify “slight” anomalies is enhanced through adversarial training, combined with POT dynamically selected thresholds to detect anomalies. We conduct ex-periments on five subsets of various open datasets, and the overall performance of CAT-POT is better than that of the baseline method. The average F1 value of CAT-POT on all datasets is 0.889, which is 5.1% higher than the suboptimal model, with the best improvement reaching 21.5%.

关键词： Training Time series analysis Quality of service Data models Real-time systems Decoding Complexity theory

来源：评论

学校读者我要写书评

暂无评论

VS-Net: A Voxel Encoding and Sparse Convolution Embedded Network for LiDAR 3D Object Detection

VS-Net: A Voxel Encoding and Sparse Convolution Embedded Net...

引用

2022 IEEE International Conference on Robotics and Biomimetics, ROBIO 2022

作者： Liu, Meng Niu, Jianwei Liu, Yu The China North Artificial Intelligence and Innovation Research Institute Beijing100072 China The China North Vehicle Research Institute Beijing100072 China The State Key Laboratory of Software Development Environment School of Computer Science and Engineering Beihang University Beijing100191 China The State Key Laboratory of Virtual Reality Technology and Systems School of Computer Science and Engineering Beihang University Beijing100191 China The Hangzhou Innovation Institute Beihang University Hangzhou310051 China The Zhengzhou University Research Institute of Industrial Technology School of Information Engineering Zhengzhou University Zhengzhou450001 China

ISBN: (纸本)9781665481090

Accurate and efficient 3D object detection in LiDAR point cloud is critical for autonomous driving. In this paper, we provide a solution for LiDAR-only detection which employs voxel features and a sparse convolution network. Specifically, we utilized voxel centers to encode point clouds' voxel features. Then, combining the sparse convolution method, we leveraged residual learning to design the backbone network. The detector output predicted results for multi-category and multi-object detection. Our method was evaluated on three popular and challenging datasets (KITTI, nuScence and Waymo). Experimental results demonstrated that the accuracy and speed (27.8 FPS) of our model could effectively support object detection for autonomous driving in various scenarios. © 2022 IEEE.

关键词： Object detection

来源：评论

学校读者我要写书评

暂无评论

Analysis and Verification of Bisimulation Relationship for Learning Time-Behavior Sequence 9

Analysis and Verification of Bisimulation Relationship for L...

引用

9th International Conference on Dependable Systems and Their Applications, DSA 2022

作者： Feng, Shu Zhu, Yi Song, Mei Gao, Yuxiang School of Computer Science and Technology Jiangsu Normal University Xuzhou China Jiangsu Education Information Engineering Technology Research Center Xuzhou China Nanjing University of Aeronautics and Astronautics Key Laboratory of Safety-Critical Software Ministry of Industry and Information Technology Nanjing China

ISBN: (数字)9781665488778

ISBN: (纸本)9781665488778

Similar learner determination has become a research hotspot in the field of personalized recommendation due to its lightweight method. At present, similar learner determination mainly adopts algorithms such as collaborative filtering, but such methods lack interpretability and cannot guarantee the reliability of the determination results. Aiming at this problem, this paper proposes a method for determining and verifying the similarity of learning time behavior sequences based on bisimulation, which uses logical reasoning to determine the similarity of learners, and uses tools to verify the correctness of the determination results. Firstly, the behavior properties of Temporal Calculus of Communication System (TCCS) are extended, and Learning Resources-Temporal Calculus of Communication System (LR-TCCS) is proposed to model the learning time-behavior sequence of learners. Secondly, we determine the similarity of learning time-behavior sequence through bisimulation. Thirdly, the bisimulation verification tool MWB is used to verify the similarity of learning time-behavior sequence, and to determine the optimal learner among similar learners through time allocation algorithm and behavior frequency analysis algorithm;Finally, the effectiveness of the method is verified by an example. © 2022 IEEE.

关键词： Calculations

来源：评论

学校读者我要写书评

暂无评论

An Intelligent Approach to Segment the Liver Cancer using Machine Learning Method 4

An Intelligent Approach to Segment the Liver Cancer using Ma...

引用

4th International Conference on Electronics and Sustainable Communication Systems, ICESC 2023

作者： Anand, L. Maurya, Mahesh Seetha, J. Nagaraju, D. Ravuri, Ananda Vidhya, R.G. School of Computing Srm Institute of Science and Technology Department of Networking and Communications India University of Mumbai K. C. College of Engineering Management and Research Mith Bunder Road Near Hume Pipe Maharashtra Thane400603 India Panimalar Engineering College Department of Computer Science and Business Systems Chennai India Sri Venkatesa Perumal College of Engineering and Technology Department of CSE Andhra Pradesh Puttur India Senior Software Engineer Intel Corporation HillsboroOR97124 United States Hkbk College of Engineering Department of Ece Bangalore India

ISBN: (纸本)9798350300093

Cancer Liver cancer is difficult to detect and diagnose early. Machine learning has improved liver cancer diagnosis in recent years. This abstract describes a novel liver cancer detection and classification method using Autoencoder-Extreme Learning Machine (AE-ELM) and Convolutional Neural Network (CNN) technology. AE-ELM models extract features in the suggested method. The autoencoder-extreme learning machine AE-ELM technique learns high-level abstract characteristics from incoming data. AE-ELM extracts liver cancer-specific features by reducing the dimensionality of input data. A CNN model classifies the extracted features. CNNs are good at detecting spatial patterns in medical pictures like CT and MRI scans. The CNN model learns hierarchical representations of input information and classifies them into liver cancer subtypes or stages using various convolutional and pooling processes. AE-ELM and CNN are advantageous. First, AE-ELM decreases input data dimensionality to extract critical features without manual feature engineering. This simplifies feature selection and improves the model's capacity to handle big data. Second, the CNN model uses hierarchical convolutional processes to capture subtle patterns and spatial correlations in input data, enhancing liver cancer classification accuracy. The proposed liver cancer diagnostic method works in experiments. AE-ELM and CNN models outperform classic machine learning approaches and standalone CNN models. Improved liver cancer detection accuracy, sensitivity, and specificity help healthcare practitioners make quick and precise diagnosis. Finally, AE-ELM and CNN technologies can identify and classify liver cancer. This unique strategy uses both models to extract relevant characteristics and capture complicated patterns, improving liver cancer diagnosis accuracy and efficiency. To evaluate this approach's therapeutic promise and application in real-world healthcare settings, additional datasets are needed. © 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

Smooth Representation Learning of Infrared and Visible Images

SSRN

引用

SSRN 2022年

Deep learning-based methods generally adopt unsupervised similarity fusion optimization mechanism (UFM) or Generative Adversarial Network-based fusion optimization mechanism (GFM) toward image fusion tasks without ground truth (GT) labels, which leads to the problem that the fused image is over smooth or under smooth. This paper adopts a simple-yet-effective self-supervised fusion representation learning optimization mechanism (FLM) via directly maximizing the mutual information between the fused image and the samples (i.e., positive and negative samples), which can be theoretically justified by its link to mutual information smoothing theory. Following InfoNCE, we further propose a smooth representation learning optimization mechanism (SFLM), which balances the fused image under smoothing and over smoothing. The former loses much high-frequency information, while the latter makes the fused image contain much harmful information. Specifically, the SFLM is optimized by minimizing the total variation loss in the pixel space and maximizing the total variation loss in the gradient space. To show the superiority of SFLM, quantitative and qualitative experiments are carried out on four public infrared and visible image fusion datasets, including RoadScene, TNO, VIFB, and MF. Extensive experiments show that our method neatly outperforms the state-of-the-art methods in most metrics. © 2022, The Authors. All rights reserved.

关键词： Image fusion

来源：评论

学校读者我要写书评

暂无评论

A Machine Learning Approach for Predicting Air Quality Index in Smart Cities

A Machine Learning Approach for Predicting Air Quality Index...

引用

IoT Based Control Networks and Intelligent Systems (ICICNIS), International Conference on

作者： S. Swamynathan N. Sneha S. P. Ramesh R. Niranjana D. David Neels Ponkumar R. Saravanakumar Department of Computer Applications Karpagam Academy of Higher Education Coimbatore Tamil Nadu India Department of Computer Science and Engineering School of Computer Science & Engineering Galgotias University Greater Noida Uttar Pradesh India Department of Electronics and Communication Engineering Sri Krishna College of Engineering and Technology Coimbatore Tamil Nadu India Department of Electronics and Communication Engineering Vel Tech Rangarajan Dr. Sagunthala R&D Institute of Science and Technology Chennai Tamil Nadu India Iconix Software Solution Tirunelveli Tamil Nadu India

ISBN: (数字)9798331518097

ISBN: (纸本)9798331518103

One essential natural resource, the air quality index has been deteriorated by economic activities. Many studies have been conducted about the forecasting of periods of unsafe atmospheric conditions; however, the majority of these studies are limited in their capacity to include seasonal and other factors due to insufficient longitudinal data. India noise pollution has compiled a 6-year dataset that has been used to build several prediction models. There are encouraging outcomes when AQI levels are predicted using machine learning methodologies such as support vector machines (SVMs) artificial neural networks (ANNs), and Naive Bayes (NB). However, in several tests involving datasets from three different locations, NB, ANN, and SVM performed well to maximize prediction performance. When it comes to mean absolute error (MAE), R-squared (R2), and Root-mean-square deviation (RMSE), the combined performance of stacks is unsurpassed. In the related work section, summarize the limitations of existing models, such as inadequate data preprocessing, low spatialtemporal resolution, limited feature selection, or overfitting in complex models. Highlight gaps in scalability, real-time adaptability, and generalization across regions, emphasizing the need for robust, efficient models tailored to air quality prediction.

关键词： Support vector machines Adaptation models Atmospheric modeling Biological system modeling Computational modeling Artificial neural networks Predictive models Air quality Data models Indexes

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：