检索结果-内蒙古大学图书馆

arXiv 2025年

作者： Liao, Zehui Hu, Shishuai Zou, Ke Fu, Huazhu Zhen, Liangli Xia, Yong National Engineering Laboratory for Integrated Aero-Space-Ground-Ocean Big Data Application Technology School of Computer Science and Engineering Northwestern Polytechnical University Xi’an710072 China Institute of High Performance Computing Agency for Science Technology and Research Singapore138632 Singapore National University of Singapore Singapore119077 Singapore

Multimodal large language models (MLLMs) have demonstrated significant potential in medical Visual Question Answering (VQA). Yet, they remain prone to hallucinations—incorrect responses that contradict input images, posing substantial risks in clinical decision-making. Detecting these hallucinations is essential for establishing trust in MLLMs among clinicians and patients, thereby enabling their real-world adoption. Current hallucination detection methods, especially semantic entropy (SE), have demonstrated promising hallucination detection capacity for LLMs. However, adapting SE to medical MLLMs by incorporating visual perturbations presents a dilemma. Weak perturbations preserve image content and ensure clinical validity, but may be overlooked by medical MLLMs, which tend to over rely on language priors. In contrast, strong perturbations can distort essential diagnostic features, compromising clinical interpretation. To address this issue, we propose Vision Amplified Semantic Entropy (VASE), which incorporates weak image transformations and amplifies the impact of visual input, to improve hallucination detection in medical VQA. We first estimate the semantic predictive distribution under weak visual transformations to preserve clinical validity, and then amplify visual influence by contrasting this distribution with that derived from a distorted image. The entropy of the resulting distribution is estimated as VASE. Experiments on two medical open-ended VQA datasets demonstrate that VASE consistently outperforms existing hallucination detection methods. Copyright © 2025, 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

Quality Assessment of Screen Content Videos

Quality Assessment of Screen Content Videos

引用

International Conference on Pattern Recognition and Image Analysis (IPRIA)

作者： Hossein Motamednia Pooryaa Cheraaqee Azadeh Mansouri Ahmad Mahmoudi-Aznaveh Department of Electrical and Computer Engineering Faculty of Engineering Kharazmi University Tehran Iran High Performance Computing Laboratory School of Computer Science Institute for Research in Fundamental Sciences Tehran Iran Cyber Research Institute Shahid Beheshti University Tehran Iran

Perceptual quality assessment has always been challenging due to the difficulty in modeling the no-linear human visual system. With the diversity in the contents of multimedia signals, the conventional methods for traditional media seems no longer satisfying. One of these emerging media, is the screen content images/videos (SCINs), Containing texts and computer generated graphics, SCVs cannot be sufficiently expressed with features designed for natural sceneries. Therefore, new researches tried to devise objective quality assessment metrics, specificly for screen contents. Recently, a dataset was proposed for quality assessment of screen content videos. Since screen contents are full of structures that spread in cardinal directions, we were motivated to employ the horizontal and vertical subbands of the wavelet transform to characterize these types of visual contents. The features were incorporated in a full-reference method that showed promising results on the publicly available dataset for SCV quality assessment. The method can bo accessed via: https://***/motamedNia/QASCV.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Harnessing Resource and Demand Flexibility for Energy Management in Urban Micro-grids 11

Harnessing Resource and Demand Flexibility for Energy Manage...

引用

11th International Conference on Innovative Smart Grid Technologies - Asia, ISGT-Asia 2022

作者： Subramanian, Lalitha Jiyan, Wu Tjandra, Rudy Troitzsch, Sebastian Massier, Tobias Teh, Erine Siew Pheng Migne, Romain Yan, Xu EDF Lab Singapore Singapore TUMCREATE Ltd. Singapore Singapore Institute of Technology Engineering Cluster Singapore Institute for High Performance Computing Agency for Science Technology and Research Singapore Singapore Institute of Technology Chemical Engineering and Food Technology Cluster Singapore School of Electrical and Electronic Engineering Nanyang Technological University Singapore

ISBN: (纸本)9798350399660

Successful adoption of distributed clean energy resources requires the enhancement of system flexibility from both the generation and the demand side. This paper presents a case study on the Punggol Digital District of Singapore exploring both distributed generation, storage, and demand side flexibility. The distributed generation resources investigated include rooftop photovoltaic systems and waste-to-electricity generation via hydrogen and fuel cell technology. In the urban use case, the local demand to be served far exceeds the local distributed clean energy generation potential. Therefore in this context, flexibility is the key added value of collective self-consumption, which could be leveraged as a valuable grid service in the city-state to support the integration of variable renewable energy resources. In this context, this work presents the assessment of the flexibility that can be provided by the cooling demand and electric vehicles by the control of respective demand-side resources. © 2022 IEEE.

关键词： Energy management

来源：评论

学校读者我要写书评

暂无评论

Self-Supervised Superpixel Correspondences For Video Object Segmentation 22

Self-Supervised Superpixel Correspondences For Video Object ...

引用

Proceedings of the 2022 4th International Conference on Video, Signal and Image Processing

作者： Clement Tan Chai Kiat Yeo Cheston Tan Basura Fernando School of Computer Science and Engineering Nanyang Technological University Singapore and Agency for Science Technology and Research (A*STAR) Singapore School of Computer Science and Engineering Nanyang Technological University Singapore Centre for Frontier AI Research Agency for Science Technology and Research (A*STAR) Singapore Institute of High Performance Computing (IHPC) Agency for Science Technology and Research (A*STAR) Singapore

ISBN: (纸本)9781450397810

Prior self-supervised video object segmentation models directly find pixel correspondences between pairs of frames. Instead, we propose a novel approach of employing superpixel features for learning visual correspondences between frames of a video sequence. With an attention mechanism, we train the model to reconstruct the next frame using superpixel features between adjacent frames of videos trained in a self-supervised manner. As superpixels are able to group pixels based on color and proximity, they reduce repetitive and noisy pixels that may confuse the model during tracking. In addition, the structural information provided by the superpixel features can help improve the segmentation of objects. We benchmark our method (S3CNet) against existing self-supervised pixel correspondence frameworks such as [12, 22, 24] and show that our self-supervised superpixel correspondence method performs better and is a viable alternative for visual tracking on the commonly-used DAVIS 2017 video segmentation benchmark.

关键词： Video Object Segmentation Self-Supervised Learning Superpixels

来源：评论

学校读者我要写书评

暂无评论

Dual-scale Enhanced and Cross-generative Consistency Learning for Semi-supervised Medical Image Segmentation

arXiv

引用

arXiv 2023年

作者： Gu, Yunqi Zhou, Tao Zhang, Yizhe Zhou, Yi He, Kelei Gong, Chen Fu, Huazhu PCA Laboratory The School of Computer Science and Engineering Nanjing University of Science and Technology Nanjing210094 China The School of Computer Science and Engineering Southeast University Nanjing211189 China The Medical School Nanjing University Nanjing210023 China The National Institute of Healthcare Data Science Nanjing University Nanjing210023 China The Institute of High Performance Computing A*STAR Singapore

Medical image segmentation plays a crucial role in computer-aided diagnosis. However, existing methods heavily rely on fully supervised training, which requires a large amount of labeled data with time-consuming pixel-wise annotations. Moreover, accurately segmenting lesions poses challenges due to variations in shape, size, and location. To address these issues, we propose a novel Dual-scale Enhanced and Cross-generative consistency learning framework for semi-supervised medical image Segmentation (DEC-Seg). First, we propose a Cross-level Feature Aggregation (CFA) module that integrates cross-level adjacent layers to enhance the feature representation ability across different resolutions. To address scale variation, we present a scale-enhanced consistency constraint, which ensures consistency in the segmentation maps generated from the same input image at different scales. This constraint helps handle variations in lesion sizes and improves the robustness of the model. Furthermore, we propose a cross-generative consistency scheme, in which the original and perturbed images can be reconstructed using cross-segmentation maps. This consistency constraint allows us to mine effective feature representations and boost the segmentation performance. To further exploit the scale information, we propose a Dual-scale Complementary Fusion (DCF) module that integrates features from two scale-specific decoders operating at different scales to help produce more accurate segmentation maps. Extensive experimental results on multiple medical segmentation tasks (polyp, skin lesion, and brain glioma) demonstrate the effectiveness of our DECSeg against other state-of-the-art semi-supervised segmentation approaches. The implementation code will be released at https://***/taozh2017/DECSeg. Copyright © 2023, The Authors. All rights reserved.

关键词： Self-supervised learning

来源：评论

学校读者我要写书评

暂无评论

Learning discretized neural networks under Ricci flow

The Journal of Machine Learning Research

引用

The Journal of Machine Learning Research 2024年第1期25卷 18901-18944页

作者： Jun Chen Hanwen Chen Mengmeng Wang Guang Dai Ivor W. Tsang Yong Liu Institute of Cyber-Systems and Control Zhejiang University China and School of Computer Science and Technology Zhejiang Normal University China Institute of Cyber-Systems and Control Zhejiang University China SGIT AI Lab State Grid Corporation of China China Centre for Frontier Artificial Intelligence Research Agency for Science Technology and Research (A*STAR) Singapore and Institute of High Performance Computing Agency for Science Technology and Research (A*STAR) Singapore and College of Computing and Data Science Nanyang Technological University Singapore

In this paper, we study Discretized Neural Networks (DNNs) composed of low-precision weights and activations, which suffer from either infinite or zero gradients due to the nondifferentiable discrete function during training. Most training-based DNNs in such scenarios employ the standard Straight-Through Estimator (STE) to approximate the gradient w.r.t. discrete values. However, the use of STE introduces the problem of gradient mismatch, arising from perturbations in the approximated gradient. To address this problem, this paper reveals that this mismatch can be interpreted as a metric perturbation in a Riemannian manifold, viewed through the lens of duality theory. Building on information geometry, we construct the Linearly Nearly Euclidean (LNE) manifold for DNNs, providing a background for addressing perturbations. By introducing a partial differential equation on metrics, i.e., the Ricci flow, we establish the dynamical stability and convergence of the LNE metric with the L2-norm perturbation. In contrast to previous perturbation theories with convergence rates in fractional powers, the metric perturbation under the Ricci ow exhibits exponential decay in the LNE manifold. Experimental results across various datasets demonstrate that our method achieves superior and more stable performance for DNNs compared to other representative training-based methods.

关键词： discretized neural networks gradient perturbation information geometry ricci flow riemannian manifold

来源：评论

学校读者我要写书评

暂无评论

Analysis Of Crosstalk in Binary Weighted Bulk Acoustic Wave Transducers For Ultrasonic Based Fourier Transform Accelerators

Analysis Of Crosstalk in Binary Weighted Bulk Acoustic Wave ...

引用

IEEE Symposium (IUS) Ultrasonics

作者： Daniel Ssu-Han Chen Shyam Trivedi Xing Haw Marvin Tan Yong Shun Teo Jaibir Sharma Zaifeng Yang Viet Phuong Bui Ching Eng Png Amit Lal Kevin Tshun Chuan Chai Technology and Research (A*Star) Institute of Microeletronics (IME) Agency for Science Singapore Technology and Research (A*Star) Institute of High Performance Computing (IHPC) Agency for Science Singapore School of Electrical and Computer Engineering Cornell University Ithaca NY USA

This paper introduces an ultrasonic wave-based analog computing accelerator for 2D-FT computation. The investigation centers on two distinct pixel cell designs within the accelerator, highlighting the potential of one design to address scalability and fabrication challenges. The study compares segregated and unified binary weighted BAW transducers as transmitter, along with a comparative assessment of coupling between the two designs through modeling, simulation and experimental analysis. The result shows minimal difference in the crosstalk for both designs, supporting the scalability of the transmitter array for enhanced resolution.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Enhancing and Adapting in the Clinic: Source-free Unsupervised Domain Adaptation for Medical Image Enhancement

arXiv

引用

arXiv 2023年

作者： Li, Heng Lin, Ziqin Qiu, Zhongxi Li, Zinan Niu, Ke Guo, Na Fu, Huazhu Hu, Yan Liu, Jiang The Research Institute of Trustworthy Autonomous Systems Southern University of Science and Technology Shenzhen China The Department of Computer Science and Engineering Southern University of Science and Technology Shenzhen China The Institute of High Performance Computing Agency for Science Technology and Research Singapore Computer School Beijing Information Science and Technology University Beijing China The School of Computer and Communication Engineering University of Science and Technology Beijing Beijing China

Medical imaging provides many valuable clues involving anatomical structure and pathological characteristics. However, image degradation is a common issue in clinical practice, which can adversely impact the observation and diagnosis by physicians and algorithms. Although extensive enhancement models have been developed, these models require a well pre-training before deployment, while failing to take advantage of the potential value of inference data after deployment. In this paper, we raise an algorithm for source-free unsupervised domain adaptive medical image enhancement (SAME), which adapts and optimizes enhancement models using test data in the inference phase. A structure-preserving enhancement network is first constructed to learn a robust source model from synthesized training data. Then a teacher-student model is initialized with the source model and conducts source-free unsupervised domain adaptation (SFUDA) by knowledge distillation with the test data. Additionally, a pseudo-label picker is developed to boost the knowledge distillation of enhancement tasks. Experiments were implemented on ten datasets from three medical image modalities to validate the advantage of the proposed algorithm, and setting analysis and ablation studies were also carried out to interpret the effectiveness of SAME. The remarkable enhancement performance and benefits for downstream tasks demonstrate the potential and generalizability of SAME. The code is available at https://***/liamheng/Annotation-free-Medical-Image-Enhancement. Copyright © 2023, The Authors. All rights reserved.

关键词： Distillation

来源：评论

学校读者我要写书评

暂无评论

A K-Nearest Centroid Neighbor with atention Classifier 5

A K-Nearest Centroid Neighbor with atention Classifier

引用

5th International Conference on computer science and Artificial Intelligence, CSAI 2021

作者： Huang, Rui Ma, Ying Wang, Tian Li, GuoQi Yan, Ming Department of Computer and Information Engineering Xiamen University of Technology China Institute of High Performance Computing Agency for Science Technology and Research Singapore Singapore School of Information Science and Engineering Hunan First Normal University China Department of Precision Instrument Center for Brain Inspired Computing Research

ISBN: (纸本)9781450384155

Among classic algorithms of data mining, the K-nearest neighbor based methods are simple and effective pattern classification algorithms. However, most KNN-based methods do not fully take into account the impact of different training sample points on classification, lead to inaccurate classification. To address this issue, we propose a scheme named Attention-based local mean K-Nearest Centroid Neighbor Classifier (ALMKNCN), combining nearest centroid neighbor with attention mechanism, the influence of each training sample on the query sample is fully considered. Given the query pattern, we first calculate the local centroid mean vector for each class, and then use the idea of attention mechanism to calculate the weight of pseudo-distance between each class and test sample. Finally, based on attention coefficient, the distances between the query sample and local mean vectors are weighted to determine the class of the query sample. Extensive experiments on UCI and KEEL data sets are carried out by comparing ALMKNCN to the state-of-art KNN-based methods. The experimental results demonstrate that the proposed ALMKNCN outperforms the related competitive KNN-based methods with more effectiveness. © 2021 ACM.

关键词： Data mining

来源：评论

学校读者我要写书评

暂无评论

Fundus Image Quality Assessment and Enhancement: a Systematic Review

arXiv

引用

arXiv 2025年

作者： Li, Heng Li, Haojin Ou, Mingyang Yu, Xiangyang Zhang, Xiaoqing Niu, Ke Fu, Huazhu Liu, Jiang Research Institute of Trustworthy Autonomous Systems SUSTech Shenzhen China Department of Computer Science and Engineering SUSTech Shenzhen China Center for High Performance Computing and Shenzhen Key Laboratory of Intelligent Bioinformatics Shenzhen Institute of Advanced Technology Chinese Academy of Sciences Shenzhen China Computer School Beijing Information Science and Technology University Beijing China Singapore

As an affordable and convenient eye scan, fundus photography holds the potential for preventing vision impairment, especially in resource-limited regions. However, fundus image degradation is common under intricate imaging environments, impacting following diagnosis and treatment. Consequently, image quality assessment (IQA) and enhancement (IQE) are essential for ensuring the clinical value and reliability of fundus images. While existing reviews offer some overview of this field, a comprehensive analysis of the interplay between IQA and IQE, along with their clinical deployment challenges, is lacking. This paper addresses this gap by providing a thorough review of fundus IQA and IQE algorithms, research advancements, and practical applications. We outline the fundamentals of the fundus photography imaging system and the associated interferences, and then systematically summarize the paradigms in fundus IQA and IQE. Furthermore, we discuss the practical challenges and solutions in deploying IQA and IQE, as well as offer insights into potential future research directions. Copyright © 2025, The Authors. All rights reserved.

关键词： Photography

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：