检索结果-内蒙古大学图书馆

IS and T International Symposium on Electronic Imaging: 21st image processing: Algorithms and Systems, IPAS 2023

作者： Hao, Guanhua Roberts, Eric J. Chavez, Tanny Zhao, Zhuowen Holman, Elizabeth A. Yanxon, Howard Green, Adam Krishnan, Harinarayan Ushizima, Daniela McReynolds, Dylan Schwarz, Nicholas Zwart, Petrus H. Hexemer, Alexander Parkinson, Dilworth Y. Lawrence Berkeley National Laboratory BerkeleyCA94720 United States Lawrence Berkeley National Laboratory BerkeleyCA94720 United States Lawrence Berkeley National Laboratory BerkeleyCA94720 United States Argonne National Laboratory LemontIL60439 United States Lawrence Berkeley National Laboratory BerkeleyCA94720 United States

Scientific user facilities present a unique set of challenges for image processing due to the large volume of data generated from experiments and simulations. Furthermore, developing and implementing algorithms for real-time processing and analysis while correcting for any artifacts or distortions in images remains a complex task, given the computational requirements of the processing algorithms. In a collaborative effort across multiple Department of Energy national laboratories, the "MLExchange"project is focused on addressing these challenges. MLExchange is a Machine Learning framework deploying interactive web interfaces to enhance and accelerate data analysis. The platform allows users to easily upload, visualize, label, and train networks. The resulting models can be deployed on real data while both results and models could be shared with the scientists. The MLExchange web-based application for image segmentation allows for training, testing, and evaluating multiple machine learning models on hand-labeled tomography data. This environment provides users with an intuitive interface for segmenting images using a variety of machine learning algorithms and deep-learning neural networks. Additionally, these tools have the potential to overcome limitations in traditional image segmentation techniques, particularly for complex and low-contrast images. © 2023, Society for Imaging Science and Technology.

关键词： image segmentation

来源：评论

学校读者我要写书评

暂无评论

A Fast View Synthesis Implementation Method for Light Field applications

引用

ACM TRANSACTIONS ON MULTIMEDIA COMPUTING COMMUNICATIONS AND applications 2021年第4期17卷 1-20页

作者： Gao, Wei Zhou, Linjie Tao, Lvfang Peking Univ Sch Elect & Comp Engn Shenzhen Grad Sch Shenzhen 518055 Peoples R China Peng Cheng Lab Artificial Intelligence Res Ctr Shenzhen 518055 Peoples R China

View synthesis (VS) for light field images is a very time-consuming task due to the great quantity of involved pixels and intensive computations, which may prevent it from the practical three-dimensional real-time systems. In this article, we propose an acceleration approach for deep learning-based light field view synthesis, which can significantly reduce calculations by using compact-resolution (CR) representation and super-resolution (SR) techniques, as well as light-weight neural networks. The proposed architecture has three cascaded neural networks, including a CR network to generate the compact representation for original input views, a VS network to synthesize new views from down-scaled compact views, and a SR network to reconstruct high-quality views with full resolution. All these networks are jointly trained with the integrated losses of CR, VS, and SR networks. Moreover, due to the redundancy of deep neural networks, we use the efficient light-weight strategy to prune filters for simplification and inference acceleration. Experimental results demonstrate that the proposed method can greatly reduce the processing time and become much more computationally efficient with competitive image quality.

关键词： Deep learning-based view synthesis real-time acceleration compact representation filter pruning deep neural network compression light field systems

来源：评论

学校读者我要写书评

暂无评论

Interpreting Convolutional networks Trained on Textual Data 10th

Interpreting Convolutional Networks Trained on Textual Data

引用

10th International Conference on Pattern Recognition applications and Methods (ICPRAM)

作者： Marzban, Reza Crick, Christopher Oklahoma State Univ Dept Comp Sci Stillwater OK 74078 USA

ISBN: (纸本)9789897584862

There have been many advances in the artificial intelligence field due to the emergence of deep learning. In almost all sub-fields, artificial neural networks have reached or exceeded human-level performance. However, most of the models are not interpretable. As a result, it is hard to trust their decisions, especially in life and death scenarios. In recent years, there has been a movement toward creating explainable artificial intelligence, but most work to date has concentrated on image processing models, as it is easier for humans to perceive visual patterns. There has been little work in other fields like natural language processing. In this paper, we train a convolutional model on textual data and analyze the global logic of the model by studying its filter values. In the end, we find the most important words in our corpus to our model's logic and remove the rest (95%). New models trained on just the 5% most important words can achieve the same performance as the original model while reducing training time by more than half. Approaches such as this will help us to understand NLP models, explain their decisions according to their word choices, and improve them by finding blind spots and biases.

关键词： Explainable artificial Intelligence Natural Language processing Deep Learning artificial neural networks Convolutional networks

来源：评论

学校读者我要写书评

暂无评论

Reversible adversarial information hiding based on the interpretation of neural network

引用

DIGITAL SIGNAL processing 2025年 165卷

作者： Li, Ruiping Wang, Xi Cui, Qingchen Li, Ming Henan Normal Univ Informat Construct & Management Off Xinxiang 453007 Peoples R China Nanyang Normal Univ Sch Artificial Intelligence & Software Engn Nanyang 473061 Peoples R China Henan Engn Res Ctr Intelligent Proc Big Data Digit Nanyang 473000 Henan Peoples R China Zhoukou Vocat Coll Arts & Sci Coll Informat Engn Zhoukou 466000 Peoples R China Henan Normal Univ Coll Comp & Informat Engn Xinxiang 453007 Peoples R China Key Lab Artificial Intelligence & Personalized Lea Xinxiang 453007 Peoples R China

The emerging technology of adversarial information hiding can generate adversarial example by embedding useful information instead of meaningless noise into the host image. The obtained image can deceive classification models and therefore equip the function of privacy protection. However, existing research mainly focuses on the effectiveness of attack rather than preserving the integrity of the protected image. It cannot meet the requirements in the fields such as medical imaging, financial transactions, and copyright protection where users' raw data should not be damaged. Therefore, we propose a novel reversible adversarial information hiding based on interpretability of neural networks. The Grad-CAM-generated heatmap is employed to identify a minimal set of high-impact pixels for embedding, ensuring that minor modifications can induce significant misclassification. Then, the user-defined secret bits are embedded into the identified pixels by using difference expansion. After extraction, the original image can be perfectly restored. The technologies of adversarial example and reversible information hiding are combined to accommodate wider applications. The proposed method is tested on CIFAR-10 with three different neural architectures (NiN, AlexNet, ResNet). Experimental results show that the proposed method can completely restore the cover image while ensuring the extraction of the embedded data and misleading the neural network.

关键词： Adversarial attacks Adversarial information hiding Reversibility neural network interpretability

来源：评论

学校读者我要写书评

暂无评论

Cocktail Glass Network: Fast Depth Estimation Using Channel to Space Unrolling

引用

IEEE ACCESS 2021年 9卷 114680-114689页

作者： Yu, Jung-Jae Ko, Jong-Gook Kim, Junmo Elect & Telecommun Res Inst ETRI Daejeon 34129 South Korea Korea Adv Inst Sci & Technol KAIST Daejeon 34141 South Korea

Depth-estimation from a single input image can be used in applications such as robotics and autonomous driving. Recently, depth-estimation networks with UNet encoder/decoder structures have been widely used. In these decoders, operations are repeated to gradually increase the image resolution, while decreasing the channel size. If the upsampling operation at a high magnification can be processed at once, the amount of computation in the decoder can be dramatically reduced. To achieve this, we propose a new network structure, i.e., a cocktail glass network. In this network, convolution layers in the decoder are reduced, and a novel fast upsampling method is used that is known as channel-to-space unrolling, which converts thick channel data into high-resolution data. The proposed method can be easily implemented using simple reshaping operations;therefore, it is suitable for reducing the depth-estimation network. Considering the experimental results based on the NYU V2 and KITTI datasets, we demonstrate that the proposed method reduces the amount of computation in the decoder by half, while maintaining the same level of accuracy;it can be used in both lightweight and large-model-capacity networks.

关键词： Decoding Convolution image resolution Three-dimensional displays Task analysis Licenses Glass artificial neural networks image processing depth estimation monocular image

来源：评论

学校读者我要写书评

暂无评论

3rd International Conference on Recent Trends in Machine Learning, IoT, Smart Cities and applications, ICMISC 2022

3rd International Conference on Recent Trends in Machine Lea...

引用

3rd International Conference on Recent Trends in Machine Learning, IoT, Smart Cities and applications, ICMISC 2022

ISBN: (纸本)9789811960871

The proceedings contain 63 papers. The special focus in this conference is on Recent Trends in Machine Learning, IoT, Smart Cities and applications. The topics include: Early-Onset Identification of Stomach Cancer Using CNN;multifold Secured Bank Application Authentication Service Using Random Visual Cryptography and Multimodal Steganography with Blockchain Technology;role of Blockchain in Health Care: A Comprehensive Study;a Novel Three Dense Layered Deep Fully Connected neural Network for Hyperspectral image Classification;research Trends in artificial Intelligence and Nature Inspired Techniques;comparative Analysis of Software Defect Prediction Using Dimensionality Reduction;an Approach to Learn Structural Similarity Between Decision Trees Using Hungarian Algorithm;design and Development of IoT-Based Intelligent Cattle Shed Management;review Paper on Technologies to Curb Noise Pollution in No Honking Zones;security Threats in Healthcare Systems—A Bibliometric Study;news Channel Debate Analysis: A Detailed Insight;disease Recognization of Plant Using Different image processing Algorithm;a Detailed Survey on Sentimental Analysis on Social Media;perceiving Correlation Among Spatiotemporal Gait Parameters and Verifying Its Relation Using Machine Learning Classification Technique Pilot Study for Indian Population;radial Basis neural Network Trained Minimum Snap Trajectory for Quadrotor;REMICARE—Medicine Intake Tracker and Healthcare Assistant;Detection of Fraudulent Credit Card Transactions in Real Time Using SparkML and Kafka;real-Time Face Detection and Face Recognition: Study of Approaches;ASIC Implementation of AI Edge Network on Chip (NoC) at 28 nm Technology Node and Its Various Timing Optimization Techniques at Each Stage;3D IC Integration Using Blockchain;Impact of COVID-19 on Import and Export of Petroleum Products and Crude Oil in India.

关键词：

来源：评论

学校读者我要写书评

暂无评论

MixSegNet: Fusing multiple mixed-supervisory signals with multiple views of networks for mixed-supervised medical image segmentation

引用

ENGINEERING applications OF artificial INTELLIGENCE 2024年 133卷

作者： Wang, Ziyang Yang, Chen Univ Oxford Dept Comp Sci Wolfson BldgParks Rd Oxford OX1 3QD England Beijing Univ Chem Technol Dept Comp Sci 15 BeiSanhuan East Rd Beijing 100029 Peoples R China

Deep learning has driven remarkable advancements in medical image segmentation. The requirement for comprehensive annotations, however, poses a significant challenge due to the labor-intensive and expensive nature of expert annotation. Addressing this challenge, we introduce a multiple mixed-supervisory signals learning (MSL) strategy, MixSegNet, that synergistically harnesses the benefits of Fully-Supervised (FSL), Weakly-Supervised (WSL), and Semi-Supervised Learning (SSL). This approach enables the utilization of various data-efficient annotations for network training, promoting efficient medical image segmentation within realistic clinical scenarios. MixSegNet concurrently trains networks with a combination of limited dense labels, a larger proportion of cost-efficient sparse labels, and unlabeled data. The networks utilized in this system comprise Vision Transformer (ViT) and Convolutional neural networks (CNN), which work together via an effective strategy including network self-ensembling and label dynamic-ensembling. This strategy adeptly handles the training challenges arising from datasets with limited or absent supervisory signals. We validated MixSegNet on a public Magnetic Resonance Imaging (MRI) cardiac segmentation benchmark dataset. It demonstrated superior performance compared to 21 other SSL or WSL baseline methods under similar labeling-cost conditions, as supported by comprehensive evaluation metrics, and slightly outperform classical FSL methods. The code for MixSegNet, all baseline methods, and the data pre-processing techniques with the datasets for different annotation situations are available at https://***/ziyangwang007/MixSegNet.

关键词： Vision transformer Medical image segmentation Mixed-supervised learning

来源：评论

学校读者我要写书评

暂无评论

AI for Space Operations: the next generation of Mission Operations for Earth Observation constellations 74

AI for Space Operations: the next generation of Mission Oper...

引用

74th International Astronautical Congress, IAC 2023

作者： Schandeler, Baptiste Lamothe, Matthieu Chouteau, Florient Future Programmes Department Airbus Defence and Space 31 rue des cosmonautes Toulouse31400 France Image and Mission Engineering Department Airbus Defence and Space 31 rue des cosmonautes Toulouse31400 France Computer Vision & Advanced Data Processing Department Airbus Defence and Space 31 rue des cosmonautes Toulouse31400 France

Earth Observation Satellite constellations are more and more requested by customers to fit their strategic needs. They offer the ability to collect large amounts of data from various places, from various sensors. They offer versatility, robustness and an undeniable strategic advantage of revisit. Classical way of operations, where every satellite, subsystems and the ground segment were operated by a team, does fit neither technically nor commercially for constellations. artificial intelligence (AI) has proven its value in multiple applications. Automated mission operations, especially in multi-missions context, have a high impact in terms of efficiency where reduction of human interactions is foreseen: routine operations, monitoring, data processing and ground segment health. Airbus has already developed automated collision avoidance maneuver (CAM), automated mission plan uplink and advanced monitoring that already allow us to operate with a limited number of operators. AI-based image production and analysis is one of the latest developments Airbus has performed: change detection and Deepzoom are now available on the market. AI is set to play a major role in the automation of future EO systems, enabled by the significant advances in Machine Learning techniques of recent years. Data-driven AI-based solutions can be developed to improve operations effectiveness. Multi-variable multi-subsystems analysis (based on neural networks) for time series can be used to predict future behaviors. Identifying a potential upcoming failure before it occurs and proposing preventive procedures to reduce the downtime is one of the leaps where the classical way of operations is outdated. The spacecraft can no longer be controlled individually by operators: the impact of on-board AI will drastically shake current CONOPS. Today, Airbus is working on artificial intelligence concepts for the next-gen EO constellations: AI on board and on ground to reach the lights-out center, where AI and a

关键词： artificial Intelligence Constellations Earth Observation Machine Learning Satellite operations

来源：评论

学校读者我要写书评

暂无评论

Enhancing FPGA-Based YOLO Object Detection: Multi-Bank Storage Optimization and Model Refinement for Real-Time applications 8

Enhancing FPGA-Based YOLO Object Detection: Multi-Bank Stora...

引用

8th International Conference on Integrated Circuits and Microsystems, ICICM 2023

作者： Ma, Shiwei Bie, Zhisong Liang, Xianchen Beijing University of Posts and Telecommunications School of Artificial Intelligence Beijing China

ISBN: (纸本)9798350318517

Object detection algorithms are widely used in image classification and detection tasks recently. Among the other algorithms, You Only Look Once (YOLO) performs better in real-time detection. Convolutional neural networks (CNNs) in the algorithm require a large amount of computation and memory access. However, researchers tend to access parameters through the external memory in accelerator design, which leads to slow processing speed and high power consumption. Therefore, we propose a Multi-Bank storage structure that flexibly combines URAM and BRAM in storing feature maps, which significantly improves the throughput. We also improve the YOLOv3-Tiny model by distillation, quantization and pruning and make it more suitable for deployment in FPGAs. These methods can significantly reduce the number of parameters in the model while ensuring almost no decrease in detection accuracy. Finally, tested on VOC 2007, our model achieves a frame rate of 67.6 FPS with a reduction in mAP of 0.38%. © 2023 IEEE.

关键词： Field programmable gate arrays (FPGA)

来源：评论

学校读者我要写书评

暂无评论

ITAT 2022 - Proceedings of the 22nd Conference Information Technologies - applications and Theory

ITAT 2022 - Proceedings of the 22nd Conference Information T...

引用

22nd Conference Information Technologies - applications and Theory, ITAT 2022

The proceedings contain 26 papers. The topics discussed include: deep transfer learning of traversability assessment for heterogeneous robots;on the incremental construction of deep rule theories;text-to-ontology mapping via natural language processing models;tams: text augmentation using most similar synonyms for SMS spam filtering;using artificial neural networks to determine ontologies most relevant to scientific texts;learning to segment from object sizes;beyond sensor data analysis: unexpected challenges in a honeybee monitoring project;image classifier with dynamic set of known classes;analysis of the semantic vector space induced by a neural language model and a corpus;on reducing automata and their normalizations;and keyphrase extraction from Slovak court decisions.

关键词：

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：