检索结果-内蒙古大学图书馆

Revisiting model-agnostic private learning: faster rates and active learning

学校读者我要写书评

暂无评论

The Journal of Machine Learning Research

The Journal of Machine Learning Research 2021年第1期22卷 11936-11979页

作者： Chong Liu Yuqing Zhu Kamalika Chaudhuri Yu-Xiang Wang Department of Computer Science University of California Santa Barbara CA Department of Computer Science and Engineering University of California San Diego La Jolla CA

The Private Aggregation of Teacher Ensembles (PATE) framework is one of the most promising recent approaches in differentially private learning. Existing theoretical analysis shows that PATE consistently learns any VC-classes in the realizable setting, but falls short in explaining its success in more general cases where the error rate of the optimal classifier is bounded away from zero. We fill in this gap by introducing the Tsybakov Noise Condition (TNC) and establish stronger and more interpretable learning bounds. These bounds provide new insights into when PATE works and improve over existing results even in the narrower realizable setting. We also investigate the compelling idea of using active learning for saving privacy budget, and empirical studies show the effectiveness of this new idea. The novel components in the proofs include a more refined analysis of the majority voting classifier -- which could be of independent interest -- and an observation that the synthetic "student" learning problem is nearly realizable by construction under the Tsybakov noise condition.

关键词： model-agnostic private learning private aggregation of teacher ensembles differential privacy Tsybakov noise condition active learning

Zea Mays Multi-Disease Classification and Severity Assessment with EfficientNetV2 Variants

学校读者我要写书评

暂无评论

Zea Mays Multi-Disease Classification and Severity Assessmen...

IEEE Region 10 Symposium

作者： Ivan Roy S. Evangelista Melvin K. Cabatuan Lorelyn Joy T. Milagrosa Argel A. Bandala Ronnie Concepcion Elmer P. Dadios Department of Electronics and Computer Engineering De La Salle University Manila Philippines Electronics Engineering Department Sultan Kudarat State University Sultan Kudarat Philippines Department of Crop Science University of Southern Mindanao North Cotabato Philippines Department of Manufacturing Engineering and Management De La Salle University Manila Philippines

Despite the advances achieved in crop protection and management, crop diseases remain a problem for corn farmers. Numerous studies have presented the efficacy of corn disease detection and classification using machine learning-based vision detectors. However, many of these models rely on datasets with lab-based images that do not depict the real-world accurately. In this study, a vision-based corn disease detector and classifier is developed with EfficientNetV2 as base architecture. In addition, an algorithm for evaluating the severity of the disease has also been created. Two datasets were built to train the models, the common corn diseases (CCD) and corn disease severity (CDS). The EfficientNetV2-B0, B1, B2, B3, and S models, pre-trained on ImageN et, were explored for feature extraction. A custom classifier head is incorporated into the EfficientNetV2-based model to complete the architecture. It consists of a single convolutional neural network (CNN) and two fully connected layers. Transfer learning and fine-tuning were employed to improve the performance. The models were evaluated based on accuracy, cross-entropy loss, precision, recall, and F1-score. The EfficientNetV2B2 model performed best on the disease classification task, with an accuracy of 95.74%. The EfficientNetV2B3 is the top performer in the disease severity assessment task, with an accuracy of 98.73%. The EfficientNetV2S also surpassed other proposed models in PlantVillage (PV) dataset with an accuracy of 99.52%.

关键词：

Prediction Model of Dioxin Emission Concentration in MSW Incineration Process Based on Stochastic Configuration Network 7

学校读者我要写书评

暂无评论

Prediction Model of Dioxin Emission Concentration in MSW Inc...

7th International Conference on Information science and Control engineering, ICISCE 2020

作者： DIng, Chenxi Yan, Aijun Wang, DIanhui Beijing University of Technology Faculty of Information Technology Beijing China La Trobe University Dept. of Computer Science and Computer Engineering Melbourne Australia

ISBN: (纸本)9781728164069

The increasing municipal solid waste has become one of the stubborn diseases of modern society. Incineration technology with the advantage of reduction and resource utilization is an important avenue to solve this problem. Toxic gases emitted from incineration not only pollute the environment but also pose a serious threat to human health. However, the complexity of incineration process brings great difficulty to the detection of pollutants concentration in the exhaust gases. In order to address the problem that the concentration of pollutants is difficult to predict, a prediction model of dioxin emission concentration based on stochastic configuration network is developed in this paper. The prediction model takes the online operation data of municipal solid waste incineration power plant as input parameters and realizes the online prediction of dioxin emission concentration. It provides a scientific basis for choosing the appropriate control strategy. When compared with BP neural network, RBF neural network and support vector machine prediction model, it is easily found that the proposed prediction model can effectively improve the prediction accuracy of dioxin emission concentration during the municipal solid waste incineration process. © 2020 IEEE.

关键词： Municipal solid waste

Personalized Hotel Recommender System Based on Graded Logic with Asymmetric Criteria

学校读者我要写书评

暂无评论

Procedia computer science 2024年 246卷 2864-2873页

作者： Andres Solano-Barliza Aïda Valls Antonio Moreno Jozo Dujmovic Melisa Acosta-Coll Jose Escorcia-Gutierrez Emiro De-la-Hoz-Franco Departament d’Enginyeria Informàtica i Matemàtiques Universitat Rovira i Virgili 43007 Tarragona Spain Department of Computational Science and Electronics Universidad de la Costa CUC 080002 Barranquilla Colombia Faculty of Engineering Universidad de La Guajira 44001 Riohacha-La Guajira Colombia Department of Computer Science San Francisco State University 94132 San Francisco CA USA

Personalizing hotel recommendations for tourists is challenging because it needs to model a logical aggregation of multiple criteria in terms of the user’s decision process and preferences. In this paper, we propose a hierarchical multi-criteria hotel evaluation model that includes important hotel properties as well as hotel ratings by previous customers. The recommender system proposed, called ***, applies different graded logic aggregation operators following the method of Logic Scoring of Preference (LSP). Asymmetric aggregation is used for some of the criteria, where the user can not only define the importance of inputs, but also indicate if they are mandatory or optional. This is the first recommender system based on graded logic and the first non-professional use of LSP. The application of graded logic operators significantly improves the decision-making process of hotel selection for each user. The case of Riohacha (an emerging tourist destination in Colombia) is studied. In addition to quality and location criteria, some security information is included in the model as it is of utmost importance to provide to travelers a secure experience at the destination.

关键词： Recommender system Decision Support Asymmetric aggregation Graded Logic Uncertainty Tourism

Generative AI Enables Medical Image Segmentation in Ultra Low-Data Regimes

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Zhang, Li Jindal, Basu Alaa, Ahmed Weinreb, Robert Wilson, David Segal, Eran Zou, James Xie, Pengtao Department of Electrical and Computer Engineering University of California La Jolla San DiegoCA United States Bakar Computational Health Sciences Institute University of California San Francisco San FranciscoCA United States Department of Electrical Engineering and Computer Sciences University of California Berkeley BerkeleyCA United States Hamilton Glaucoma Center Shiley Eye Institute Viterbi Family Department of Ophthalmology University of California La Jolla San DiegoCA United States Division of Pulmonary Allergy and Critical Care Medicine Department of Medicine University of Pittsburgh PittsburghPA United States Department of Computer Science and Applied Mathematics Weizmann Institute of Science Rehovot Israel Department of Molecular Cell Biology Weizmann Institute of Science Rehovot Israel Department of Biomedical Data Science Stanford University School of Medicine StanfordCA United States Department of Computer Science Stanford University StanfordCA United States Department of Medicine University of California La Jolla San DiegoCA United States

Semantic segmentation of medical images is pivotal in applications like disease diagnosis and treatment planning. While deep learning has excelled in automating this task, a major hurdle is the need for numerous annotated segmentation masks, which are resource-intensive to produce due to the required expertise and time. This scenario often leads to ultra low-data regimes, where annotated images are extremely limited, posing significant challenges for the generalization of conventional deep learning methods on test images. To address this, we introduce a generative deep learning framework, which uniquely generates high-quality paired segmentation masks and medical images, serving as auxiliary data for training robust models in data-scarce environments. Unlike traditional generative models that treat data generation and segmentation model training as separate processes, our method employs multi-level optimization for end-to-end data generation. This approach allows segmentation performance to directly influence the data generation process, ensuring that the generated data is specifically tailored to enhance the performance of the segmentation model. Our method demonstrated strong generalization performance across 9 diverse medical image segmentation tasks and on 16 datasets, in ultra-low data regimes, spanning various diseases, organs, and imaging modalities. When applied to various segmentation models, it achieved performance improvements of 10-20% (absolute), in both same-domain and out-of-domain scenarios. Notably, it requires 8 to 20 times less training data than existing methods to achieve comparable results. This advancement significantly improves the feasibility and cost-effectiveness of applying deep learning in medical imaging, particularly in scenarios with limited data availability. Copyright © 2024, The Authors. All rights reserved.

关键词： Semantic Segmentation

UMS: Live Migration of Containerized Services across Autonomous Computing Systems

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Chanikaphon, Thanawat Salehi, Mohsen Amini HPCC Lab School of Computing and Informatics University of Louisiana LafayetteLA United States HPCC Lab Computer Science and Engineering Department University of North Texas United States

Containerized services deployed within various computing systems, such as edge and cloud, desire live migration support to enable user mobility, elasticity, and load balancing. To enable such a ubiquitous and efficient service migration, a live migration solution needs to handle circumstances where users have various authority levels (full control, limited control, or no control) over the underlying computing systems. Supporting the live migration at these levels serves as the cornerstone of interoperability, and can unlock several use cases across various forms of distributed systems. As such, in this study, we develop a ubiquitous migration solution (called UMS) that, for a given containerized service, can automatically identify the feasible migration approach, and then seamlessly perform the migration across autonomous computing systems. UMS does not interfere with the way the orchestrator handles containers and can coordinate the migration without the orchestrator involvement. Moreover, UMS is orchestrator-agnostic, i.e., it can be plugged into any underlying orchestrator platform. UMS is equipped with novel methods that can coordinate and perform the live migration at the orchestrator, container, and service levels. Experimental results show that for single-process containers, the service-level approach, and for multi-process containers with small ( Copyright © 2023, The Authors. All rights reserved.

关键词： Containers

Deep Learning based Defect classification and detection in SEM images: A Mask R-CNN approach

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Dey, Bappaditya Dehaerne, Enrique Khalil, Kasem Halder, Sandip Leray, Philippe Bayoumi, Magdy A. Imec Kapeldreef 75 Leuven3001 Belgium Department of Electrical and Computer Engineering University of Mississippi Mississippi United States The Center for Advanced Computer Studies University of Louisiana LafayetteLA United States Department of Electrical Engineering Assiut University Egypt Department of Electrical and Computer Engineering University of Louisiana LafayetteLA United States Department of Computer Science KU Leuven Leuven Belgium

In this research work, we have demonstrated the application of Mask-RCNN (Regional Convolutional Neural Network), a deep-learning algorithm for computer vision and specifically object detection, to semiconductor defect inspection domain. Stochastic defect detection and classification during semiconductor manufacturing has grown to be a challenging task as we continuously shrink circuit pattern dimensions (e.g., for pitches less than 32 nm). Defect inspection and analysis by state-of-the-art optical and e-beam inspection tools is generally driven by some rule-based techniques, which in turn often causes to misclassification and thereby necessitating human expert intervention. In this work, we have revisited and extended our previous deep learning-based defect classification and detection method towards improved defect instance segmentation in SEM images with precise extent of defect as well as generating a mask for each defect category/instance. This also enables to extract and calibrate each segmented mask and quantify the pixels that make up each mask, which in turn enables us to count each categorical defect instances as well as to calculate the surface area in terms of pixels. We are aiming at detecting and segmenting different types of inter-class stochastic defect patterns such as bridge, break, and line collapse as well as to differentiate accurately between intra-class multi-categorical defect bridge scenarios (as thin/single/multi-line/horizontal/non-horizontal) for aggressive pitches as well as thin resists (High NA applications). Our proposed approach demonstrates its effectiveness both quantitatively and qualitatively. Copyright © 2022, The Authors. All rights reserved.

关键词： Stochastic systems

Multi-target and multi-stage liver lesion segmentation and detection in multi-phase computed tomography scans

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Al-Battal, Abdullah F. Duong, Soan T.M. Tang, Van Ha Tran, Quang Duc Truong, Steven Q.H. Phan, Chien Nguyen, Truong Q. An, Cheolhong Electrical and Computer Engineering Department UC San Diego La Jolla CA United States Electrical Engineering Department King Fahd University of Petroleum and Minerals Saudi Arabia Vinbrain JSC Hanoi Viet Nam Department of Computer Science Le Quy Don Technical University Hanoi Viet Nam Diagnostic Imaging Department University Medical Center Ho Chi Minh City Viet Nam

Multi-phase computed tomography (CT) scans use contrast agents to highlight different anatomical structures within the body to improve the probability of identifying and detecting anatomical structures of interest and abnormalities such as liver lesions. Yet, detecting these lesions remains a challenging task as these lesions vary significantly in their size, shape, texture, and contrast with respect to surrounding tissue. Therefore, radiologists need to have an extensive experience to be able to identify and detect these lesions. Segmentation-based neural networks can assist radiologists with this task. Current state-of-the-art lesion segmentation networks use the encoder-decoder design paradigm based on the UNet architecture where the multi-phase CT scan volume is fed to the network as a multi-channel input. Although this approach utilizes information from all the phases and outperform single-phase segmentation networks, we demonstrate that their performance is not optimal and can be further improved by incorporating the learning from models trained on each single-phase individually. Our approach comprises three stages. The first stage identifies the regions within the liver where there might be lesions at three different scales (4, 8, and 16 mm). The second stage includes the main segmentation model trained using all the phases as well as a segmentation model trained on each of the phases individually. The third stage uses the multi-phase CT volumes together with the predictions from each of the segmentation models to generate the final segmentation map. To demonstrate its ability to generalize beyond liver lesions, we also test our approach on brain lesions in multi-contrast magnetic resonance imaging (MRI). Overall, our approach improves relative liver lesion segmentation performance by 1.6% while reducing performance variability across subjects by 8% when compared to the current state-of-the-art models. © 2024, CC BY-NC-ND.

关键词： Convolutional neural networks

Net Separation-Oriented Printed Circuit Board Placement via Margin Maximization

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Cheng, Chung-Kuan Ho, Chia-Tung Holtz, Chester Department of Computer Science and Engineering University of California La Jolla San DiegoCA92037 United States

Packaging has become a crucial process due to the paradigm shift of More than Moore. Addressing manufacturing and yield issues is a significant challenge for modern layout algorithms. We propose to use printed circuit board (PCB) placement as a benchmark for the packaging problem. A maximum-margin formulation is devised to improve the separation between nets. Our framework includes seed layout proposals, a coordinate descent-based procedure to optimize routability, and a mixed-integer linear programming method to legalize the layout. We perform an extensive study with 14 PCB designs and an open-source router. We show that the placements produced by NS-place improve routed wirelength by up to 25%, reduce the number of vias by up to 50%, and reduce the number of DRVs by 79% compared to manual and wirelength-minimal placements. Copyright © 2022, The Authors. All rights reserved.

关键词： Printed circuit boards