检索结果-内蒙古大学图书馆

arXiv 2024年

作者： Hwang, Uiwon Lee, Jonghyun Shin, Juhyeon Yoon, Sungroh Division of Digital Healthcare Yonsei University Korea Republic of Department of Electrical and Computer Engineering Seoul National University Korea Republic of Interdisciplinary Program in Artificial Intelligence Seoul National University Korea Republic of

In the face of the deep learning model’s vulnerability to domain shift, source-free domain adaptation (SFDA) methods have been proposed to adapt models to new, unseen target domains without requiring access to source domain data. Although the potential benefits of applying data augmentation to SFDA are attractive, several challenges arise such as the dependence on prior knowledge of class-preserving transformations and the increase in memory and computational requirements. In this paper, we propose Source-free Domain Adaptation Through the Lens of Data Augmentation (SF(DA)2), a novel approach that leverages the benefits of data augmentation without suffering from these challenges. We construct an augmentation graph in the feature space of the pretrained model using the neighbor relationships between target features and propose spectral neighborhood clustering to identify partitions in the prediction space. Furthermore, we propose implicit feature augmentation and feature disentanglement as regularization loss functions that effectively utilize class semantic information within the feature space. These regularizers simulate the inclusion of an unlimited number of augmented target features into the augmentation graph while minimizing computational and memory demands. Our method shows superior adaptation performance in SFDA scenarios, including 2D image and 3D point cloud datasets and a highly imbalanced dataset. Copyright © 2024, 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

Accumulating Curriculum Knowledge Based on Neural Network Isolated Algorithm

Accumulating Curriculum Knowledge Based on Neural Network Is...

引用

IEEE International Symposium on Industrial Electronics (ISIE)

作者： Cheng-Fu Liou Yi-Cheng Yu Meta Zen Artificial Intelligence Research Labs (MZAIR Labs) Tainan Taiwan Graduate Degree Program of College of Electrical and Computer Engineering National Yang Ming Chiao Tung University Hsinchu Taiwan

ISBN: (数字)9798350394085

ISBN: (纸本)9798350394092

Feeding the training data to neural networks in a specific sequence, from the simple data to the difficult data, utilizing curriculum learning can enhance performance improvements over the typical learning strategy based on shuffling training samples randomly, without any additional computational costs. This training approach has been successfully applied in all fields of machine learning. However, the current curriculum learning will gradually introduce difficult samples until the model eventually performs joint training. In addition, our experimental results find and demonstrate how the order of training tasks arranged according to the difficulty of different tasks influences the reuse percentage of model capacity, allowing us to observe the impact of curriculum learning on model performance from different perspectives. Finally, we also separately conduct three ablations analyses of our model to comprehensively enhance the understanding of the impacts of the proposed approach and further demonstrate its scalability to a standard large-scale dataset, i.e., the ImageNet dataset.

关键词： Training Industrial electronics Scalability Knowledge based systems Training data Machine learning Multitasking

来源：评论

学校读者我要写书评

暂无评论

Picosecond optical-to-x-ray (Cd,Mg)Te radiation detectors

Picosecond optical-to-x-ray (Cd,Mg)Te radiation detectors

引用

2023 Conference on Lasers and Electro-Optics, CLEO 2023

作者： Cheng, J. Komissarov, I. Chen, G. Chakraborty, D. Kutcher, S. Chen, H. Trivedi, S. Sobolewski, Roman Materials Science Graduate Program University of Rochester RochesterNY14627 United States Laboratory for Laser Energetics University of Rochester RochesterNY14623 United States Department of Electrical and Computer Engineering University of Rochester RochesterNY14627 United States Brimrose Technology Corp SparksMD21152 United States

We developed a dual optical/x-ray ultrafast photodetector based on in-house grown Cdo * Mg0.03Te single crystals. The detector is characterized by ~200 ps full-width-at-half-maximum, readout-electronics limited photor... 详细信息

ISBN: (纸本)9781957171258

关键词： X ray detectors

来源：评论

学校读者我要写书评

暂无评论

On mitigating stability-plasticity dilemma in CLIP-guided image morphing via geodesic distillation loss

arXiv

引用

arXiv 2024年

作者： Oh, Yeongtak Lee, Saehyung Hwang, Uiwon Yoon, Sungroh Department of Electrical and Computer Engineering Seoul National University Korea Republic of Interdisciplinary Program in Artificial Intelligence Seoul National University Korea Republic of Division of Digital Healthcare Yonsei University Korea Republic of

Large-scale language-vision pre-training models, such as CLIP, have achieved remarkable text-guided image morphing results by leveraging several unconditional generative models. However, existing CLIP-guided image morphing methods encounter difficulties when morphing photorealistic images. Specifically, existing guidance fails to provide detailed explanations of the morphing regions within the image, leading to misguidance. In this paper, we observed that such misguidance could be effectively mitigated by simply using a proper regularization loss. Our approach comprises two key components: 1) a geodesic cosine similarity loss that minimizes inter-modality features (i.e., image and text) on a projected subspace of CLIP space, and 2) a latent regularization loss that minimizes intra-modality features (i.e., image and image) on the image manifold. By replacing the naïve directional CLIP loss in a drop-in replacement manner, our method achieves superior morphing results on both images and videos for various benchmarks, including CLIP-inversion. © 2024, CC BY.

关键词： Distillation

来源：评论

学校读者我要写书评

暂无评论

ProPILE: probing privacy leakage in large language models 23

ProPILE: probing privacy leakage in large language models

引用

Proceedings of the 37th International Conference on Neural Information Processing Systems

作者： Siwon Kim Sangdoo Yun Hwaran Lee Martin Gubri Sungroh Yoon Seong Joon Oh Department of Electrical and Computer Engineering Seoul National University NAVER AI Lab University of Luxembourg and Parameter Lab Department of Electrical and Computer Engineering Seoul National University and Interdisciplinary Program in Artificial Intelligence Seoul National University Parameter Lab and Tübingen AI Center University of Tübingen

The rapid advancement and widespread use of large language models (LLMs) have raised significant concerns regarding the potential leakage of personally identifiable information (PII). These models are often trained on vast quantities of web-collected data, which may inadvertently include sensitive personal data. This paper presents ProPILE, a novel probing tool designed to empower data subjects, or the owners of the PII, with awareness of potential PII leakage in LLM-based services. ProPILE lets data subjects formulate prompts based on their own PII to evaluate the level of privacy intrusion in LLMs. We demonstrate its application on the OPT-1.3B model trained on the publicly available Pile dataset. We show how hypothetical data subjects may assess the likelihood of their PII being included in the Pile dataset being revealed. ProPILE can also be leveraged by LLM service providers to effectively evaluate their own levels of PII leakage with more powerful prompts specifically tuned for their in-house models. This tool represents a pioneering step towards empowering the data subjects for their awareness and control over their own data on the web. The demo can be found here: https://***/research/propile

关键词：

来源：评论

学校读者我要写书评

暂无评论

Multivariate Time-Series Deep Learning for Joint Prediction of Temperature and Relative Humidity in a Closed Space

引用

Procedia computer Science 2023年 227卷 1046-1053页

作者： Fergianto E. Gunawan Arief S. Budiman Bens Pardamean Endang Juana Sugiarto Romeli Tjeng W. Cenggoro Kartika Purwandari Alam A. Hidayat Anak. A.N.P. Redi Muhammad Asrol Industrial Engineering Department BINUS Graduate Program – Master of Industrial Engineering Bina Nusantara University Jakarta 11480 Indonesia Computer Science Department BINUS Graduate Program – Master of Computer Science Program Bina Nusantara University Jakarta 11480 Indonesia Bioinformatics and Data Science Research Center Bina Nusantara University Jakarta 11480 Indonesia Electrical Engineering Department Faculty of Industrial Technology Universitas Trisakti Jakarta 11440 Indonesia PT Impack Pratama Industri Jakarta 14350 Indonesia Computer Science Department School of Computer Science Bina Nusantara University Jakarta 11480 Indonesia Mathematics Department School of Computer Science Bina Nusantara University Jakarta 11480 Indonesia

An accurate predictive model of temperature and humidity plays a vital role in many industrial processes that utilize a closed space such as in agriculture and building management. With the exceptional performance of deep learning on time-series data, developing a predictive temperature and humidity model with deep learning is propitious. In this study, we demonstrated that deep learning models with multivariate time-series data produce remarkable performance for temperature and relative humidity prediction in a closed space. In detail, all deep learning models that we developed in this study achieve almost perfect performance with an R value over 0.99.

关键词： closed space deep learning humity prediction indoor temperature prediction

来源：评论

学校读者我要写书评

暂无评论

Effects of Incident Angle on Single-Event Leakage Current Degradation in Heavy-Ion Irradiated 1800 v and 4500 v SiC Power MOSFETs

引用

IEEE Transactions on Nuclear Science 2025年

作者： Sengupta, A. Martinella, C. Ball, D.R. Niskanen, K. Senarath, A.S. Islam, S. Harris, P.M. Hutson, J.M. Osheroff, J.M. Sternberg, A.L. Kettunen, H. Kosier, S.L. Grossner, U. Javanainen, A. Galloway, K.F. Schrimpf, R.D. Witulski, A.F. Zürich8092 Switzerland Vanderbilt University Interdisciplinary Materials Science Program NashvilleTN37235 United States Lipscomb University Electrical and Computer Engineering Department NashvilleTN37204 United States NASA/GSFC Code 561.4 GreenbeltMD20771 United States Vanderbilt University Electrical Engineering and Computer Science Department B 351825 NashvilleTN37235 United States University of Jyväskylä Department of Physics Jyväskylä40014 Finland

Experimental heavy-ion responses of two variants of SiC power MOSFETs are evaluated. The devices have similar epitaxial thickness but different doping. The higher doping in the epitaxial region results in lower breakdown voltage, and typically leads to a lower threshold drain voltage at which single-event leakage current (SELC) occurs. Furthermore, the heavy-ion-induced SELC degradation in the drain and gate currents of the SiC MOSFETs exhibit a strong dependence on the heavy-ion's angle of incidence. The devices' behavior significantly alters when irradiated with heavy-ions at different incident angles under the same fixed bias and total fluence, as also confirmed by 3-D Technology computer Aided Design (TCAD) simulations. © 1963-2012 IEEE.

关键词： Power MOSFET

来源：评论

学校读者我要写书评

暂无评论

Semantic Communications for Artificial Intelligence Generated Content (AIGC) Toward Effective Content Creation

arXiv

引用

arXiv 2023年

作者： Liu, Guangyuan Du, Hongyang Niyato, Dusit Kang, Jiawen Xiong, Zehui Kim, Dong In Shen, Xuemin The School of Computer Science and Engineering The Energy Research Institute @ NTU Interdisciplinary Graduate Program Nanyang Technological University Singapore The School of Computer Science and Engineering Nanyang Technological University Singapore The School of Automation Guangdong University of Technology China The Pillar of Information Systems Technology and Design Singapore University of Technology and Design Singapore The Department of Electrical and Computer Engineering Sungkyunkwan University Korea Republic of The Department of Electrical and Computer Engineering University of Waterloo Canada

Artificial Intelligence Generated Content (AIGC) Services have significant potential in digital content creation. The distinctive abilities of AIGC, such as content generation based on minimal input, hold huge potential, especially when integrating with semantic communication (SemCom). In this paper, a novel comprehensive conceptual model for the integration of AIGC and SemCom is developed. Particularly, a content generation level is introduced on top of the semantic level that provides a clear outline of how AIGC and SemCom interact with each other to produce meaningful and effective content. Moreover, a novel framework that employs AIGC technology is proposed as an encoder and decoder for semantic information, considering the joint optimization of semantic extraction and evaluation metrics tailored to AIGC services. The framework can adapt to different types of content generated, the required quality and the semantic information utilized. By employing a Deep Q Network (DQN), a case study is presented that provides useful insights into the feasibility of the optimization problem and its convergence characteristics. © 2023, CC BY.

关键词： Wireless networks

来源：评论

学校读者我要写书评

暂无评论

Variance-stationary Differentiable NAS 32

Variance-stationary Differentiable NAS

引用

32nd British Machine Vision Conference, BMVC 2021

作者： Choe, Hyeokjun Na, Byunggook Mok, Jisoo Yoon, Sungroh Department of Electrical and Computer Engineering Seoul National University Seoul Korea Republic of Interdisciplinary Program in AI Seoul National University Seoul Korea Republic of

Differentiable architecture search (DARTS) has become the popular method of neural architecture search (NAS) due to its adaptability and low computational cost. However, following the publication of DARTS, it has been found that DARTS often yields a suboptimal neural architecture because architecture parameters do not accurately represent operation strengths. Through extensive theoretical analysis and empirical observations, we reveal that this issue occurs as a result of the existence of unnormalized operations. Based on our finding, we propose a novel variance-stationary differentiable architecture search (VS-DARTS), which consists of node normalization, local adaptive learning rate, and pβ-continuous relaxation. Comprehensively, VS-DARTS makes the architecture parameters a more reliable metric for deriving a desirable architecture without increasing the search cost. In addition to the theoretical motivation behind all components of VS-DARTS, we provide strong experimental results to demonstrate that they synergize to significantly improve the search performance. The architecture searched by VS-DARTS achieves the test error of 2.50% on CIFAR-10 and 24.7% on ImageNet. © 2021. The copyright of this document resides with its authors.

关键词： computer vision

来源：评论

学校读者我要写书评

暂无评论

A BCLC Staging System for Hepatocellular Carcinoma using Swin Transformer and CT Imaging 46

A BCLC Staging System for Hepatocellular Carcinoma using Swi...

引用

46th Annual International Conference of the IEEE engineering in Medicine and Biology Society, EMBC 2024

作者： Chang, Shun-Cheng Wang, Pochuang Wang, Weichung Su, Tung-Hung Kao, Jia-Horng Lin, Che Taipei10617 Taiwan Department of Computer Science and Information Engineering Ntu Taipei10617 Taiwan Institute of Applied Mathematical Sciences Ntu Taipei10617 Taiwan Division of Gastroenterology and Hepatology Department of Internal Medicine National Taiwan University Hospital Taipei10048 Taiwan Department of Electrical Engineering Ntu Taipei10617 Taiwan Center for Advanced Computing and Imaging in Biomedicine Ntu Taipei10617 Taiwan Smart Medicine and Health Informatics Program Ntu Taipei10617 Taiwan

ISBN: (纸本)9798350371499

The Barcelona Clinic Liver Cancer (BCLC) staging system plays a crucial role in clinical planning, offering valuable insights for effectively managing hepatocellular carcinoma. Accurate prediction of BCLC stages can significantly ease the workload on radiologists. However, few datasets are explicitly designed for discerning BCLC stages. Despite the common practice of appending BCLC labels to clinical data within datasets, the inherent imbalance in BCLC distribution is further amplified by the diverse purposes for which datasets are curated. In this study, we aim to develop a BCLC staging system using the advanced Swin Transformer model. Additionally, we explore the integration of two datasets, each originally intended for separate objectives, highlighting the critical challenge of preserving class distribution in practical study designs. This exploration is pivotal for ensuring the applicability of our developed staging system in the designed clinical settings. Our resulting BCLC staging system demonstrates an accuracy of 55.81% (±7.8%), contributing to advancing medical image-based research for predicting BCLC stages. © 2024 IEEE.

关键词： Distribution transformers

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：