检索结果-内蒙古大学图书馆

5th IEEE International conference on Artificial Intelligence and Big Data, ICAIBD 2022

作者： Chen, Xuehui Jiang, Jingfei Niu, Xin Pan, Hengyue Dong, Peijie Wei, Zimian National University of Defense Technology National Laboratory for Parallel and Distributed Processing Changsha China

ISBN: (纸本)9781665499132

Multi-objective neural architecture search (NAS) algorithms aim to automatically search the neural architecture suitable for different computing power platforms by using multi-objective optimization methods. The LEMONADE algorithm, which is a representative algorithm of multi-objective NAS algorithms, maintains a population of networks on an approximation of the Pareto front of the multiple objectives, such as predictive performance, number of parameters or FLOPs. To address the irrationality and repeatability of only sampling based on cheap objectives in LEMONADE, we propose a novel multi-objective neural architecture search algorithm via Gaussian Process sampling, dubbed GP-LEMONADE. Meanwhile, to make the sampling process more efficient, we design the online predictor based on Gaussian Process to predict expensive objectives, and sample candidate networks by combining cheap objectives and expensive objectives, so as to ensure the rationality and efficiency of sampling. Experiments show that the GP-LEMONADE algorithm evolves 100 generations and obtains the SOTA model with 3.98% test error. This process only takes 7.38 GPU days, which is 26.75 GPU days shorter than that of LEMONADE. Our methods have improved the performance of the LEMONADE algorithm and ensured the rationality and efficiency of sampling during the evolution, which effectively improving the search efficiency of multi-objective NAS algorithms. © 2022 IEEE.

关键词： Multiobjective optimization

来源：评论

学校读者我要写书评

暂无评论

Performance evaluation of a many/multi-core implementation for signal filtering problem 16

Performance evaluation of a many/multi-core implementation f...

引用

16th International conference on Signal-image Technology and Internet-Based Systems (SITIS)

作者： D'Alo, Vincenzo De Luca, Pasquale Parthenope Univ Naples Sci & Technol Dept Naples Italy

ISBN: (纸本)9781665464956

Signal filtering plays an important role in pre-processing procedures used in several application fields. In last years, several methods have been proposed by achieving good results. When large data are given in input, performance are likely to get worse. Based on the Savitzky-Golay filter, we have developed, in this paper, a parallel implementation with engaging a multi-core architecture. Experiments and tests highlight the gain of performance in time terms.

关键词： parallel algorithm Savitzky-Golay signal filtering multi-core architecture performance analysis

来源：评论

学校读者我要写书评

暂无评论

Deep learning-based computer aided diagnosis model for skin cancer detection and classification

引用

distributed AND parallel DATABASES 2022年第4期40卷 717-736页

作者： Adla, Devakishan Reddy, G. Venkata Rami Nayak, Padmalaya Karuna, G. JNTUH Hyderabad India VJIT Dept IT Hyderabad India JNTUH Sch Informat Technol Hyderabad India GRIET Dept CSE Hyderabad India JNTUH Dept CSE GRIET Hyderabad India

Skin cancer is a commonly occurring disease, which affects people of all age groups. Automated detection of skin cancer is needed to decrease the death rate by identifying the diseases at the initial stage. The visual inspection during the medical examination of skin lesions is a tedious process as the resemblance among the lesions exists. Recently, imaging-based Computer Aided Diagnosis (CAD) model is widely used to screen and detect the skin cancer. This paper is designed with automated Deep Learning with a class attention layer based CAD model for skin lesion detection and classification known as DLCAL-SLDC. The goal of the DLCAL-SLDC model is to detect and classify the different types of skin cancer using dermoscopic images. During image pre-processing, Dull razor approach-based hair removal and average median filtering-based noise removal processes take place. Tsallis entropy based segmentation technique is applied to detect the affected lesion areas in the dermoscopic images. Also, a DLCAL based feature extractor is used for extracting the features from the segmented lesions using Capsule Network (CapsNet) along with CAL and Adagrad optimizer. The CAL layer incorporated into the CapsNet is intended to capture the discriminative class-specific features to cover the class dependencies and effectively bridge the CapsNet for further process. Finally, the classification is carried out by the Swallow Swarm Optimization (SSO) algorithm based Convolutional Sparse Autoencoder (CSAE) known as SSO-CSAE model. The proposed DLCAL-SLDC technique is validated using a benchmark ISIC dataset. The proposed framework has accomplished promising results with 98.50% accuracy, 94.5% sensitivity and 99.1% specificity over the other methods interms of different measures.

关键词： Skin lesion Dermoscopic images image classification ISIC dataset Deep learning image segmentation Hair removal Capsule Networks

来源：评论

学校读者我要写书评

暂无评论

How to Fit the SCC Algorithm Efficiently into distributed Graph Iterative Computation

How to Fit the SCC Algorithm Efficiently into Distributed Gr...

引用

IEEE Annual International Computer Software and Applications conference (COMPSAC)

作者： Xiaochen Sun Wei Wang Tao Huang State Key Laboratory of Computer Science Institute of Software Chinese Academy of Sciences Beijing China University of Chinese Academy of Sciences Beijing China

ISBN: (数字)9798350376968

ISBN: (纸本)9798350376975

This paper reviews the sequential, parallel and distributed implementations of strongly connected component algorithms, and analyzes the challenges of each implementation in the graph iteration paradigm of distributed processing. We also review the graph data layout and communication mode of each distributed graph processing system, and analyze the defect of high memory usage in the implementation of the strongly connected component algorithm of the existing distributed graph processing system. Therefore, we propose a strongly connected component algorithm that performs pull mode communication on CSR instead of traversing the transposed graph. Experiments show that the memory consumption of our method is greatly reduced, and the running time of the algorithm is much lower than that of the most advanced implementation. On four publicly accessible data sets, our approach reduces computation time and storage space by an average of 26% and 39%, respectively. In addition, we optimize the general pull communication mode, resulting in 15.2°/0 computation time reduction on the four publicly available data sets.

关键词： distributed processing Reviews image color analysis Layout Memory management distributed databases Iterative algorithms

来源：评论

学校读者我要写书评

暂无评论

Single Node Acceleration of Generative Adversarial Networks using HPC for image Analytics 5

Single Node Acceleration of Generative Adversarial Networks ...

引用

5th International conference on Computational Intelligence and Intelligent Systems, CIIS 2022

作者： Ravikumar, Aswathy Sriraman, Harini School of Computer Science and Engineering Vellore Institute of Technology Chennai600127 India

ISBN: (纸本)9781450397612

Generative Adversarial Networks (GAN) are approaches that are utilized for data augmentation, which facilitates the development of more accurate detection models for unusual or unbalanced datasets. Computer-assisted diagnostic methods may be made more reliable by using synthetic pictures generated by GAN. Generative adversarial networks are challenging to train because too unpredictable training dynamics may occur throughout the learning process, such as model collapse and vanishing gradients. For accurate and faster results the GAN network need to trained in parallel and distributed manner. We enhance the speed and precision of the Deep Convolutional Generative Adversarial Networks (DCGAN) architecture by using its parallelism and executing it on High-Performance Computing platforms. The effective analysis of a DCGAN in Graphic processing Unit and Tensor processing Unit platforms in which each layer execution pattern is analyzed. The bottleneck is identified for the GAN structure for each execution platforms. The Central processing Unit is capable of processing neural network models, but it requires a great deal of time to do it. Graphic processing Unit in contrast, side, are a hundred times quicker than CPUs for Neural Networks, however, they are prohibitively expensive compared to CPUs. Using the systolic array structure, TPU performs well on neural networks with high batch sizes but in GAN the shift between CPU and TPU is huge so it does not perform well. © 2022 ACM.

关键词： Graphics processing unit

来源：评论

学校读者我要写书评

暂无评论

Spatial-Frequency Integration Network with Dual Prompt Learning for Few-shot image Classification

Spatial-Frequency Integration Network with Dual Prompt Learn...

引用

International Symposium on parallel and distributed processing with Applications, ISPA

作者： Yi Liu Jie He Shoukun Xu Guangqi Jiang Academy of Computing and Artificial Intelligence Changzhou University Changzhou 213164 Jiangsu China Changzhou China

ISBN: (数字)9798331509712

ISBN: (纸本)9798331509729

Few-shot image classification is a challenging task that aims to recognize image classes based on only a few training images. However, existing methods face the following two main challenges: (1) Ignoring the frequency domain information during image feature extraction. (2) It does not take the semantic gap between multiple modalities into consideration, which limits the classification performance. To overcome these limitations, we propose a novel method named Spatial-Frequency Integration Network with Dual Prompt Learning for few-shot image classification. Firstly, we introduce a spatial-frequency integration module that combines spatial domain and low-frequency information to extract discriminative image features from the image modality. Secondly, we design a dual prompting module, which integrates learnable prompts and hand-crafted prompts to improve the generalization of applications to new classes. Thirdly, we propose an image-text interaction module to enhance inter-modal complementary and consistency. Both theoretical and experimental validations confirm the effectiveness of the proposed method in few-shot image classification.

关键词： Training image segmentation image recognition Frequency-domain analysis Face recognition Semantics Object detection Feature extraction Few shot learning image classification

来源：评论

学校读者我要写书评

暂无评论

Research on Point Cloud Data processing in Multiple Scenarios

Research on Point Cloud Data Processing in Multiple Scenario...

引用

Information Science, parallel and distributed Systems (ISPDS), International conference on

作者： Jinkui Huang Xiangyao Ma Jingyuan Wang Jianhua Li College of Communication Engineering Army Engineering University of PLA Nanjing China

ISBN: (数字)9798350352719

ISBN: (纸本)9798350352726

With the development of 3D laser scanning technology, point cloud data has been widely used in various fields. As the most critical link after the point cloud data collection, the denoising process of point cloud data directly affects the subsequent point cloud data splicing, 3D reconstruction and other links. At present, there are still gaps in the research of point cloud data processing methods under different application scenarios. Therefore, this paper studies the statistical filtering and radius filtering methods of point cloud data in different scenarios and environments, and combines the characteristics of the two filtering methods to obtain a point cloud data filtering method suitable for different scenarios.

关键词： Point cloud compression Information science Three-dimensional displays Splicing Oceans Noise reduction Noise

来源：评论

学校读者我要写书评

暂无评论

Research on Change Detection Algorithms for image Interaction and Edge Enhancement

Research on Change Detection Algorithms for Image Interactio...

引用

Information Science, parallel and distributed Systems (ISPDS), International conference on

作者： Qi Wang Jiying Li School of Electronic and Information Engineering Lanzhou Jiaotong University Lan Zhou China

ISBN: (数字)9798350352719

ISBN: (纸本)9798350352726

Most of the existing detection methods ignore the spatial and temporal characteristics of the two images and global dependence of the interconnections, resulting in many irrelevant changes (such as illumination, the atmosphere, seasonal changes in vegetation) to produce the phenomenon of misdetection; at the same time, the existing methods of change detection. Meanwhile, the existing change detection methods are not easy to obtain structural details in high-resolution remote sensing images and the detection accuracy needs to be improved. Based on the global information interaction and edge detection method of dual-time-phase image, a network model is proposed, which consists of three parts and is divided into three modules, namely, edge structure and image feature extraction, image transformer and change region discrimination. Edge detection is performed on the dual-temporal phase image to obtain the edge map, edge detection can effectively improve the edge details of the detection results, the edge map and the original image are fused for feature extraction, and the extracted dual-temporal phase features are sent to the image transformer BIT for global spatial-temporal modeling to establish the connection, modeling the remote context in the dual-temporal image in an efficient and effective way, and finally through the simple region of change discriminative module for discrimination obtaining the detection results. The algorithm achieves 96.2% F1 on CDD dataset and 93.8% F1 on BCDD dataset, which is superior with better stability and robustness compared with some known algorithms.

关键词： Change detection algorithms image edge detection Atmospheric modeling Vegetation mapping Lighting Feature extraction Transformers

来源：评论

学校读者我要写书评

暂无评论

Artificial Neural Network for processing Fingerprint image Noise 23

Artificial Neural Network for Processing Fingerprint Image N...

引用

23rd ACIS International Summer Virtual conference on Software Engineering, Artificial Intelligence, Networking and parallel/distributed Computing, SNPD-Summer 2022

作者： Han, Ke Institute of Forensic Science Ministry of Public Security China Department of Fingerprint Technology Beijing China

ISBN: (纸本)9798350396379

Fingerprint images from crime scenes can be used to find and identify suspects in the field of forensic science. Fingerprint images are usually polluted by noise, which affects the visual effect of fingerprint images. It is important to process the noise of fingerprint images. A supervised learning 9-layer artificial neural network besides the input layer and the output layer is designed to remove the noise of fingerprint images in this paper. The first three layers of the neural network are convolutional layers. The purpose of designing convolutional layers is to extract the feature information of the images layer by layer and gradually generate the feature maps. The fourth layer to the sixth layer are fully connected layers. The number of neurons in the fourth layer is the same as that of the sixth layer. The number of neurons in the fourth layer is the number of elements in the vector flattened by the feature maps outputted from the last convolutional layer. The seventh layer to the ninth layer in the neural network are deconvolutional layers. These deconvolutional layers gradually restore the feature maps to the fingerprint image. Finally, the enhanced fingerprint image is outputted from the output layer. As the training samples of fingerprint images is inadequate, this paper provides a way to make training samples. The clear fingerprint image is blurred as the corresponding blurred fingerprint image, which solves the problem of lack of the training samples. The blurred fingerprint image and the corresponding clear fingerprint image are used to establish the training sample set and train the artificial neural network. The mean square error function is used as the loss function to adjust and optimize the parameters of the neural network. The neural network in this paper is evaluated on test fingerprint images. The neural network method is compared with the Gaussian low-pass filtering method and the Wiener filtering method. Experimental results show that the neura

关键词： image enhancement

来源：评论

学校读者我要写书评

暂无评论

A Referenced Framework on New Challenges and Cutting-Edge Research Trends for Big-Data processing Using Machine Learning Approaches

A Referenced Framework on New Challenges and Cutting-Edge Re...

引用

2023 International conference on Smart Computing and Application, ICSCA 2023

作者： Kirola, Madhu Memoria, Minakshi Shuaib, Mohammed Joshi, Kapil Alam, Shadab Alshanketi, Faisal Uttaranchal University Department of Computer Engineering Dehradun India College of Computer Science & It Jazan University Jazan Saudi Arabia

ISBN: (纸本)9798350347050

Machine Learning and Big-Data Analytics is each growing computing fields. Big data is, in reality, gaining attraction in all regions of Science and engineering. While those big-data units have the clean capacity, successfully comprehending them will call for new thinking methods and inventive learning methodologies to cope with the severe issues. On the other hand, traditional machine learning strategies may not be able to fulfil the needs of real-time data processing for big-datasets because of the big-data age approaches. As a result, machine learning will reinvent itself for the big-data era. Where improvements in any discipline Big-Data analytics or machine Learning complement the opposite. This article offers a literature evaluation of the most recent advancement in ML (machine learning) approaches for large amounts of data processing. We focused on evaluating and analyzing the challenging situation and potential solutions associated with handling huge data sets with machine learning. This review article initially presents an overview of big-data and big-data analytics, followed by reviewing traditional machine-learning approaches as well as examining new aspects of machine-learning techniques for big-data processing. After that identifying the critical challenges of big-data processing correlated with the properties of big data 5Vs-volume, velocity, variety, veracity, and value, finally proposed a reference framework for big-data processing along with new emerging tools and technologies like Hadoop, distributed computing, and parallel computing along with the identification of open issues and new research trends associated with the processing of large amounts of data sets (Big-Data). © 2023 IEEE.

关键词： Data Analytics

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：