检索结果-内蒙古大学图书馆

2024 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2024

作者： Zheng, Fang Zhou, Hang Li, Guoliang Ni, Fuchuan Huazhong Agricultural University Engineering Research Center of Intelligent Technology for Agriculture Ministry of Education Key Laboratory of Smart Farming Technology for Agricultural Animals Ministry of Agriculture and Rural Affairs College of Informatics Wuhan China Huazhong Agricultural University College of Informatics Wuhan China Hubei Engineering Technology Research Center of Agricultural Big Data Engineering Research Center of Intelligent Technology for Agriculture Ministry of Education Wuhan China

ISBN: (纸本)9798350386226

While convolutional neural network(CNN) have historically achieved significant success in 3D medical image segmentation, recent advancements in Vision Transformer (ViT) have led to increased popularity of Transformer architectures in this field. Leveraging the global information extraction capabilities of Transformers, we propose a novel parallel structure combining CNN and Transformer for multimodal brain tumor segmentation, named PTCNet. We introduce two key improvements to enhance multimodal segmentation:1) Introducing a novel channel attention module that replaces spatial self-attention in the Transformer to improve cross-modal information extraction.2) Enhancing the reversible feature extraction module based on residual structures to better detect small lesions. We evaluated our method on the BraTS 2023 brain metastasis dataset, achieving an average DSC of 61.42%, which is 1.12% higher than the second place. The innovative structure and improved modules of PTCNet significantly enhance performance in multimodal brain tumor segmentation tasks. Code and models are available at https://***/naiqizh/PTCNet. © 2024 IEEE.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Attribution rollout: a new way to interpret visual transformer

引用

Journal of Ambient Intelligence and Humanized Computing 2023年第1期14卷 163-173页

作者： Xu, Li Yan, Xin Ding, Weiyue Liu, Zechao College of Computer Science and Technology Harbin Engineering University Nantong Street Heilongjiang Harbin150001 China Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education Jilin University Qianjin Street Jilin Changchun130012 China Department of Medicine Harvard Medical School Longwood Avenue BostonMA02115 United States

Transformer-based models are dominating the field of natural language processing and are becoming increasingly popular in the field of computer vision. However, the black box characteristics of transformers seriously hamper their application in certain fields. Prior work relies on the raw attention scores or employs heuristic propagation along with the attention graph. In this work, we propose a new way to visualize model. The method computes attention scores based on attribution and then propagates these attention scores through the layers. This propagation involves attention layers and multi-head attention mechanism. Our method extracts salient dependencies in each layer to visualize prediction results. We benchmark our method on recent visual transformer networks and demonstrate its many advantages over the existing interpretability methods. Our code is available at: https://***/yxheartipp/attr-rollout. © 2022, The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature.

关键词： Natural language processing systems

来源：评论

学校读者我要写书评

暂无评论

Dual-level Mixup for Graph Few-shot Learning with Fewer Tasks 25

Dual-level Mixup for Graph Few-shot Learning with Fewer Task...

引用

34th ACM Web Conference, WWW 2025

作者： Liu, Yonghao Li, Mengyu Giunchiglia, Fausto Huang, Lan Li, Ximing Feng, Xiaoyue Guan, Renchu College of Computer Science and Technology Jilin University Changchun China Department of Information Engineering and Computer Science University of Trento Trento Italy Key Laboratory of Symbolic Computation and Knowledge Engineering of the Ministry of Education China

ISBN: (纸本)9798400712746

Graph neural networks have been demonstrated as a powerful paradigm for effectively learning graph-structured data on the web and mining content from it. Current leading graph models require a large number of labeled samples for training, which unavoidably leads to overfitting in few-shot scenarios. Recent research has sought to alleviate this issue by simultaneously leveraging graph learning and meta-learning paradigms. However, these graph meta-learning models assume the availability of numerous meta-training tasks to learn transferable meta-knowledge. Such assumption may not be feasible in the real world due to the difficulty of constructing tasks and the substantial costs involved. Therefore, we propose a SiMple yet effectIve approach for graph few-shot Learning with fEwer tasks, named SMILE. We introduce a dual-level mixup strategy, encompassing both within-task and across-task mixup, to simultaneously enrich the available nodes and tasks in meta-learning. Moreover, we explicitly leverage the prior information provided by the node degrees in the graph to encode expressive node representations. Theoretically, we demonstrate that SMILE can enhance the model generalization ability. Empirically, SMILE consistently outperforms other competitive models by a large margin across all evaluated datasets with in-domain and cross-domain settings. Our anonymous code can be found here. © 2025 Copyright held by the owner/author(s).

关键词： Graph neural networks

来源：评论

学校读者我要写书评

暂无评论

PriFU: Capturing Task-Relevant Information Without Adversarial Learning 24

PriFU: Capturing Task-Relevant Information Without Adversari...

引用

32nd ACM International Conference on Multimedia, MM 2024

作者： Bi, Xiuli Hu, Yang Liu, Bo Li, Weisheng Cosman, Pamela Xiao, Bin Chongqing Key Laboratory of Image Cognition Key Laboratory of Cyberspace Big Data Intelligent Security Ministry of Education Chongqing University of Posts and Telecommunications Chongqing China Department of Electrical and Computer Engineering University of California San Diego San Diego United States

ISBN: (纸本)9798400706868

As machine learning advances, machine learning as a service (MLaaS) in the cloud brings convenience to human lives but also privacy risks, as powerful neural networks used for generation, classification or other tasks can also become privacy snoopers. This motivates privacy preservation in the inference phase. Many approaches for preserving privacy in the inference phase introduce multi-objective functions, training models to remove specific private information from users' uploaded data. Although effective, these adversarial learning-based approaches suffer not only from convergence difficulties, but also from limited generalization beyond the specific privacy for which they are trained. To address these issues, we propose a method for privacy preservation in the inference phase by removing task-irrelevant information, which requires no knowledge of the privacy attacks nor introduction of adversarial learning. Specifically, we introduce a metric to distinguish task-irrelevant information from task-relevant information, and achieve more efficient metric estimation to remove task-irrelevant features. The experiments demonstrate the potential of our method in several tasks. Our code will be available at: https://***/iwhoyoung/PriFU. © 2024 ACM.

关键词： Adversarial machine learning

来源：评论

学校读者我要写书评

暂无评论

Segment-Adaptive Spread Spectrum Audio Watermarking Using NSGA-Ⅱ

引用

Chinese Journal of Electronics 2021年第4期30卷 727-735页

作者： SU Zhaopin ZHANG Guofu ZHOU Xianxian LI Wangwang Key Laboratory of Knowledge Engineering with Big Data Hefei University of Technology Ministry of Education School of Computer Science and Information Engineering Hefei University of Technology Intelligent Interconnected Systems Laboratory of Anhui Province Hefei University of Technology Anhui Province Key Laboratory of Industry Safety and Emergency Technology Hefei University of Technology

In the field of robust audio watermarking,how to seek a good trade-off between robustness and imperceptibility is challenging. The existing studies use the same embedding parameter for each part of the audio signal, which ignores that different parts may have different requirements for embedding parameters. In this work, the constraints on imperceptibility are first ***, we present a segment multi-objective optimization model of the scaling parameter under the constrained Signal-to-noise ratio(SNR) in Spread spectrum(SS)audio watermarking. Additionally, we adopt the Nondominated sorting genetic algorithm Ⅱ(NSGA-Ⅱ) to solve the proposed model. Finally, we compare our algorithm(called SS-SNR-NSGA-Ⅱ) with the existing methods. The experimental results show that the proposed SS-SNRNSGA-Ⅱ not only provides flexible choices for different application demands but also achieves more and better trade-offs between imperceptibility and robustness.

关键词： Spread spectrum audio watermarking Scaling parameter Segment Constrained SNR NSGA-Ⅱ

来源：评论

学校读者我要写书评

暂无评论

Dual decoder UNet with Contrastive learning for Brain Image Registration

Dual decoder UNet with Contrastive learning for Brain Image ...

引用

2024 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2024

作者： Zhang, Ying Guo, Shuai Shi, Dantong Xiang, Jinhai Huazhong Agricultural University College of Informatics Wuhan430070 China Agricultural Bioinformatics Key Laboratory of Hubei Province China Key Laboratory of Smart Farming for Agricultural Animals Ministry of Agriculture Hubei Engineering Technology Research Center of Agricultural Big Data Engineering Research Center of Intelligent Technology for Agriculture Ministry of Education Huazhong Agricultural University Wuhan430070 China

ISBN: (纸本)9798350386226

The core of the image registration task is to accurately extract and compare the spatial feature information between moving and fixed images. The model must not only be able to capture features inside an image, but also be able to distinguish similarities and differences between moving image and fixed image. Although UNet and its variant networks are widely used in the field of registration, they still present challenges in terms of fine-grained and multi-level feature extraction. In order to address these issues, we propose a dual-decoder UNet network based on contrastive learning. For enhancing the discriminant ability of the network at the feature level, the proposed model applies contrastive learning to the encoder stage of UNet, which strengthens the differential learning in the process of feature extraction. In addition, a multi-level feature fusion method is introduced in the encoder to consider the feature differences at each level more comprehensively. Experimental results show that the proposed method has achieved a significant improvement in the registration accuracy. On the LPBA40 dataset, compared with Affine, VoxelMorph and CLMorph, the Dice scores were increased by 15.2%, 2.5% and 1.9%, respectively, which verified the effectiveness of the proposed method. Our code implementation is publicly available at https://***/LaneAzur/DD-CLUNet. © 2024 IEEE.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

SS-MAF: Semi-Supervised Echocardiographic Segmentation Using Multi-Atlas Fusion 8

SS-MAF: Semi-Supervised Echocardiographic Segmentation Using...

引用

8th International Conference on Communication, Image and Signal Processing, CCISP 2023

作者： Yang, Shuaihua Zheng, Caixia College of Information Science and Technology Northeast Normal University Changchun130117 China Jilin University Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education Changchun130012 China

ISBN: (纸本)9798350305838

Left ventricle (LV) segmentation in echocardiography is of paramount significance in cardiac function analysis. Recently, semi-supervised segmentation has garnered considerable attention due to its potential mitigation of labeling costs. Despite this, most semi-supervised methodologies tend to overlook the specialized anatomical priors specific to echocardiography, often yielding suboptimal outcomes. Simultaneously, atlas priors have been validated on other organs and imaging modalities. However, their direct application in echocardiography is fraught with challenges, primarily attributed to the indistinct boundaries of the LV and the presence of fan-shaped image edges, which impede atlas image registration. This paper presents a new method of semi-supervised segmentation using multi-atlas fusion, named SS- MAF, which is designed to incorporate atlas priors tailored for echocardiography while circumventing the intricate image registration process in traditional atlas-based methods. The SS-MAF comprises two components: a Multi-Task Learning (MTL) module and a Multi-Atlas Fusion (MAF) module. The MTL module serves precise landmark localization and segmentation estimation. The MAF module is primarily focused on the generation of probabilistic atlases, utilizing annotated atlas data with the guidance of key landmarks to facilitate the automatic annotation of unlabeled data, avoiding the traditional image-level registration, thus enhancing the performance of the MTL module. Our SS- MAF is rigorously evaluated on the publicly available CAMUS dataset, with results demonstrating its superiority over other state-of-the-art semi-supervised segmentation techniques in echocardiography. © 2023 IEEE.

关键词： Image registration

来源：评论

学校读者我要写书评

暂无评论

Assessing the Potential Tidal Energy Resources in Coastal Areas of Fujian Province 6

Assessing the Potential Tidal Energy Resources in Coastal Ar...

引用

6th International Conference on Energy Systems and Electrical Power, ICESEP 2024

作者： Jiala, Erge Wu, Wenting Fuzhou University Key Laboratory of Spatial Data Mining and Information Sharing of Ministry of Education National & Local Joint Engineering Research Center of Satellite Geospatial Information Technology The Academy of Digital China Fuzhou China

ISBN: (纸本)9798350353563

In the context of increasingly serious depletion and security risks of fossil energy, environmental pollution, and challenges of climate change, tidal range energy, and tidal current energy have attracted considerable attention due to their high predictability, insensitivity to flood and drought periods, and minimal environmental impact. In this study, we aim to assess the status of tidal range energy and tidal current energy in the coastal areas of Fujian Province using a high-resolution tidal dynamic model developed from the state-of-the-art model, Delft3D FM. Our results indicate that tidal range energy resources are concentrated in the coastal port areas north of central Fujian, with the highest tidal range energy potential of 1.18×105 KWh/m2 in Xinghua Bay. Tidal range energy resources are primarily concentrated in the waters of Sansha Bay, Pingtan Island, and Quanzhou Bay. Among them, the potential annual average power in Sansha Bay can reach 2.60×105 kWh/m2, with an estimated exploitable capacity of 67.24MW after the installation of tidal current power stations at the inlet channels. These research findings provide data support for site selection in the development of tidal range energy resources in the offshore waters of Fujian. © 2024 IEEE.

关键词： Delft3D FM High-resolution modelling Potential resources assessment Tidal energy

来源：评论

学校读者我要写书评

暂无评论

SGSH: Stimulate Large Language Models with Skeleton Heuristics for knowledge Base Question Generation

arXiv

引用

arXiv 2024年

作者： Guo, Shasha Liao, Lizi Zhang, Jing Wang, Yanling Li, Cuiping Chen, Hong School of Information Renmin University of China Beijing China Key Laboratory of Data Engineering and Knowledge Engineering of Ministry of Education China Singapore Management University Singapore Zhongguancun Laboratory China

knowledge base question generation (KBQG) aims to generate natural language questions from a set of triplet facts extracted from KB. Existing methods have significantly boosted the performance of KBQG via pre-trained language models (PLMs) thanks to the richly endowed semantic knowledge. With the advance of pre-training techniques, large language models (LLMs) (e.g., GPT-3.5) undoubtedly possess much more semantic knowledge. Therefore, how to effectively organize and exploit the abundant knowledge for KBQG becomes the focus of our study. In this work, we propose SGSH — a simple and effective framework to Stimulate GPT-3.5 with Skeleton Heuristics to enhance KBQG. The framework incorporates "skeleton heuristics", which provides more fine-grained guidance associated with each input to stimulate LLMs to generate optimal questions, encompassing essential elements like the question phrase and the auxiliary verb. More specifically, we devise an automatic data construction strategy leveraging ChatGPT to construct a skeleton training dataset, based on which we employ a soft prompting approach to train a BART model dedicated to generating the skeleton associated with each input. Subsequently, skeleton heuristics are encoded into the prompt to incentivize GPT-3.5 to generate desired questions. Extensive experiments demonstrate that SGSH derives the new state-of-the-art performance on the KBQG tasks. The code is now available on Github1 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

Learning Stable Task-Level Manifold for Few-Shot Learning

Learning Stable Task-Level Manifold for Few-Shot Learning

引用

2023 International Conference on Digital Image Computing: Techniques and Applications, DICTA 2023

作者： Yu, Siyi Luo, Wei Li, Gang Yang, Bo Jilin University College of Computer Science and Technology China Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education China Deakin University School of Information Technology Geelong Australia

ISBN: (纸本)9798350382204

Few-shot learning (FSL) aims to learn to new concepts based on very limited data. One of the main challenges in FSL is the use of pretrained embeddings whose dimension is too high for the small sample size. While the large dimensionality helps capture diverse features in the training example, not all the features are relevant to the task at hand. Such task-irrelevant features from pre-trained embeddings can negatively impact the performance of the FSL model. To address this challenge, we propose a simple yet effective way to learn task-level manifolds that explicitly eliminate the negative effect of nuisance features. One key novelty of our approach is to address class imbalance in the training data by injecting supervised information through key virtual samples, which we named Manifold Support Points. Empirically, we demonstrate that the stability of the learnt task-level manifolds and their effectiveness in challenging tasks in FSL. Our approach shows competitive performance on several benchmark datasets, including tieredImageNet and CUB. © 2023 IEEE.

关键词： Embeddings

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

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

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：