检索结果-内蒙古大学图书馆

Contour-Aware Equipotential Learning for Semantic Segmentation

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Yin, Xu Min, Dongbo Huo, Yuchi Yoon, Sung-Eui The School of Computing Korea Advanced Institute of Science and Technology Daejeon34141 Korea Republic of The Faculty of the Department of Computer Science and Engineering Ewha Womans University Seoul03760 Korea Republic of The State Key Lab of Cad and Cg Zhejiang University China Zhejiang Lab 310058 China The Faculty of School of Computing Korea Advanced Institute of Science and Technology Deajeon34141 Korea Republic of

With increasing demands for high-quality semantic segmentation in the industry, hard-distinguishing semantic boundaries have posed a significant threat to existing solutions. Inspired by real-life experience, i.e., combining varied observations contributes to higher visual recognition confidence, we present the equipotential learning (EPL) method. This novel module transfers the predicted/ground-truth semantic labels to a self-defined potential domain to learn and infer decision boundaries along customized directions. The conversion to the potential domain is implemented via a lightweight differentiable anisotropic convolution without incurring any parameter overhead. Besides, the designed two loss functions, the point loss and the equipotential line loss implement anisotropic field regression and category-level contour learning, respectively, enhancing prediction consistencies in the inter/intra-class boundary areas. More importantly, EPL is agnostic to network architectures, and thus it can be plugged into most existing segmentation models. This paper is the first attempt to address the boundary segmentation problem with field regression and contour learning. Meaningful performance improvements on Pascal Voc 2012 and Cityscapes demonstrate that the proposed EPL module can benefit the off-the-shelf fully convolutional network models when recognizing semantic boundary areas. Besides, intensive comparisons and analysis show the favorable merits of EPL for distinguishing semantically-similar and irregular-shaped categories. © 2022, CC BY-NC-ND.

关键词： Semantic Segmentation

SVM-ERFE: An Enhanced Method for Identification of Metastatic and Relapsing Nature of Breast Cancer

学校读者我要写书评

暂无评论

SVM-ERFE: An Enhanced Method for Identification of Metastati...

Technological Advancements in Computational sciences (ICTACS), International Conference on

作者： Abhijit Kumar Sumit Kumar Kevin Riou Amitaya Choudhury Gurinder Singh Sudhanshu Tripathi School of Computer Science University of Petroleum and Energy Studies Dehradun India Thapar Institute of Engineering and Technology Patiala Punjab India Polytech Nantes University of Nantes France Pundit Deendayal Energy University Ahmadabad Gujrat India Group Vice-Chancellor Amity University Uttar Pradesh India Department of IT and Engineering Amity University Tashkent Uzbekistan

ISBN: (数字)9798350387490

ISBN: (纸本)9798350387506

Classifying breast cancer into metastatic non-metastatic is crucial for a women that undergoing for surgery, because of the heavy treatment associated with metastatic cancers. With the advancement of micro array technology, identifying gene signatures from large genes dataset becomes an efficient approach to predict or classify cancerous genes. We proposed an algorithm based on support vector machine using enhanced recursive feature extraction mechanism (SVM_ ERFE). The proposed algorithm identifies most significant genes signatures and thus prognosis of metastatic/relapsing nature among breast cancer patients. In order to enhance the accuracy, we kept allow to varying the number of genes after first filtering. Several experiments has performed over metastatic-295 breast cancer patient's datasets and relaps-GSE2034 dataset, and achieved the best accuracy 100%±0.0 for 158 and 307 gene signatures after first filtering respectively. Further, the model has validated using 10-fold cross validation using evaluation metrics accuracy, TPR, TNR. Moreover, proposed model can increase the dortor's confidence into the results.

关键词： Support vector machines Accuracy Filtering Computational modeling Feature extraction Prediction algorithms Breast cancer Vectors Classification algorithms Prognostics and health management

MULTIMODAL SELF-SUPERVISED LEARNING FOR LESION LOCALIZATION

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Yang, Hao Zhou, Hong-Yu Li, Cheng Huang, Weijian Liu, Jiarun Liang, Yong Shi, Guangming Zheng, Hairong Liu, Qiegen Wang, Shanshan Paul C. Lauterbur Research Center for Biomedical Imaging Shenzhen Institute of Advanced Technology Chinese Academy of Sciences Shenzhen China Peng Cheng Laboratory Shenzhen China University of Chinese Academy of Sciences Beijing China Department of Computer Science The University of Hong Kong Pok Fu Lam Hong Kong Guangzhou China Department of Electronic Information Engineering Nanchang University China

Multimodal deep learning utilizing imaging and diagnostic reports has made impressive progress in the field of medical imaging diagnostics, demonstrating a particularly strong capability for auxiliary diagnosis in cases where sufficient annotation information is lacking. Nonetheless, localizing diseases accurately without detailed positional annotations remains a challenge. Although existing methods have attempted to utilize local information to achieve fine-grained semantic alignment, their capability in extracting the fine-grained semantics of the comprehensive context within reports is limited. To address this problem, a new method is introduced that takes full sentences from textual reports as the basic units for local semantic alignment. This approach combines chest Xray images with their corresponding textual reports, performing contrastive learning at both global and local levels. The leading results obtained by this method on multiple datasets confirm its efficacy in the task of lesion localization. Copyright © 2024, The Authors. All rights reserved.

关键词： Self-supervised learning

Coherent Two-photon Backscattering and Induced Angular Quantum Correlations in Multiple-Scattered Two-Photon States of the Light

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Estakhri, Nooshin M. Norris, Theodore B. Fowler School of Engineering Chapman University OrangeCA92866 United States Institute for Quantum Studies Chapman University OrangeCA92866 United States Department of Electrical Engineering and Computer Science University of Michigan Ann ArborMI48109 United States The Gérard Mourou Center for Ultrafast Optical Science University of Michigan Ann ArborMI48109 United States

We present the emergence of coherent two-photon backscattering, a manifestation of weak localization, in multiple scattering of maximally entangled pure and fully mixed two-photon states and examine the effect of entanglement and classical correlations. Quantum correlations in backscattering are investigated for finite three-dimensional disordered structures in the weak localization regime as well as systems of a small number of scatterers with specified spatial arrangements. No assumptions are made on the statistical behavior of the scattering matrix elements. Furthermore, we study the interplay between quantum correlations induced by multiple scattering and the correlations that may be present in the illumination fields, and how they are manifested in the output modes. We study the effect of the dimensionality of the entanglement and the angular distribution of the jointly measurable photon pairs on the emergence of enhancement and angular quantum correlations and show how quantum correlations can be used as a probe of the entanglement dimensionality. We show that by increasing the disordered material density, the width of the coherent two-photon backscattering cones increases, in accordance with the reduction of the mean free path length within the structure. Copyright © 2024, The Authors. All rights reserved.

关键词： Multiple scattering

AlphaChimp: Tracking and Behavior Recognition of Chimpanzees

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Ma, Xiaoxuan Lin, Yutang Xu, Yuan Kaufhold, Stephan P. Terwilliger, Jack Meza, Andres Zhu, Yixin Rossano, Federico Wang, Yizhou Center on Frontiers of Computing Studies School of Computer Science Peking University China Yuanpei College Peking University China Department of Cognitive Science University of California San Diego United States Institute for Artificial Intelligence Peking University China Nat'l Eng. Research Center of Visual Technology Nat'l Key Lab of General Artificial Intelligence Peking University China

Understanding non-human primate behavior is crucial for improving animal welfare, modeling social behavior, and gaining insights into both distinctly human and shared behaviors. Despite recent advances in computer vision, automated analysis of primate behavior remains challenging due to the complexity of their social interactions and the lack of specialized algorithms. Existing methods often struggle with the nuanced behaviors and frequent occlusions characteristic of primate social dynamics. This study aims to develop an effective method for automated detection, tracking, and recognition of chimpanzee behaviors in video footage. Here we show that our proposed method, AlphaChimp, an end-to-end approach that simultaneously detects chimpanzee positions and estimates behavior categories from videos, significantly outperforms existing methods in behavior recognition. AlphaChimp achieves approximately 10% higher tracking accuracy and a 20% improvement in behavior recognition compared to state-of-the-art methods, particularly excelling in the recognition of social behaviors. This superior performance stems from AlphaChimp's innovative architecture, which integrates temporal feature fusion with a Transformer-based self-attention mechanism, enabling more effective capture and interpretation of complex social interactions among chimpanzees. Our approach bridges the gap between computer vision and primatology, enhancing technical capabilities and deepening our understanding of primate communication and sociality. We release our code and models at project page and hope this will facilitate future research in animal social dynamics. This work contributes to ethology, cognitive science, and artificial intelligence, offering new perspectives on social intelligence. Copyright © 2024, The Authors. All rights reserved.

关键词： Mammals

Tipping Prediction of a Class of Large-Scale Radial-Ring Neural Networks

学校读者我要写书评

暂无评论

SSRN

SSRN 2024年

作者： Lu, Yunxiang Xiao, Min Wu, Xiaoqun Karimi, Hamid Reza Xie, Xiangpeng Cao, Jinde Zheng, Wei Xing College of Automation College of Artificial Intelligence Nanjing University of Posts and Telecommunications Nanjing210023 China College of Computer Science and Software Engineering Shen Zhen University Shen Zhen518060 China Department of Mechanical Engineering Politecnico di Milano Milan20156 Italy Institute of Advanced Technology Nanjing University of Posts and Telecommunications Nanjing210023 China School of Mathematics Southeast University Nanjing210096 China School of Computer Data and Mathematical Sciences Western Sydney University SydneyNSW2751 Australia

Understanding the emergence of collective dynamics in large-scale neural networks remains a challenging endeavor. This paper aims to address this gap by utilizing dynamic systems theory, particularly emphasizing tipping mechanisms. First of all, we introduce a novel $(n+mn)$-scale radial-ring neural network and utilize the topological approach of Coates' flow graph to determine the characteristic equation of the linearized network. Secondly, by deriving stability conditions and predicting the tipping point using the algebraic approach based on the integral element idea, we identify significant factors such as synaptic transmission delay, self-feedback coefficient, and network topology. Finally, we validate the effectiveness of the methodology in predicting the tipping point through numerical simulations. The simulations provide a thorough portrayal of the dynamics exhibited by large-scale neural networks, while also integrating robustness tests. This research contributes to a deeper understanding of the mechanisms underlying collective dynamics in large-scale neural networks, offering valuable insights for both theoretical frameworks and practical applications. © 2024, The Authors. All rights reserved.

关键词： Flow graphs

Revisiting the Topology of Urban Road Networks with Graph Embeddings

学校读者我要写书评

暂无评论

Revisiting the Topology of Urban Road Networks with Graph Em...

Chinese Automation Congress (CAC)

作者： Kang Liu Feng Wang Bing He Wei Wang Guohua Wu Jincheng Jiang Diping Yuan Shenzhen Institute of Advanced Technology Chinese Academy of Sciences Shenzhen China Intelligent Science & Technology Academy Limited Limited of CASIC Beijing China School of Computer Science Chengdu University of Information Technology Chengdu China Department of Mechanical Engineering City University of Hong Kong Hongkong China Shenzhen Technology University Shenzhen China School of Safety Engineering China University of Mining and Technology Xuzhou China

ISBN: (数字)9798350368604

ISBN: (纸本)9798350368611

Representing and depicting topological structures of road networks is beneficial for route planning, traffic management and network optimization. Even though road segments are basic management units in practice, the topological structures of road-segment based networks have not been well characterized due to the homogeneity of their network metrics (e.g., degree). Therefore, this study introduced graph-embedding technique and investigated its potential for characterizing topological structures of road-segment based networks. A popular graph-embedding model called Node2Vec is applied to learn embeddings of road segments from the networks. On this basis, we firstly explored the relationship between road-segment embeddings and centrality metrics by testing if the latter can be predicted by the former. Second, we investigated the clustering results of the road-segment embeddings and tested if the clustered road segments are modularized in topology. Experiments based on road networks of six Chinese cities proved that road-segment embeddings can not only well represent various centrality metrics (e.g., degree, closeness, and PageRank), but also detect community structures that road segments with similar embeddings tend to be densely connected with each other. Those findings revealed the characteristics of road embeddings and proved their potential for representing road-segment based network structures, thus we can better apply them to more related downstream tasks such as transport network optimization and traffic prediction.

关键词： Measurement Automation Network topology Roads Urban areas Topology Planning Optimization Testing

DNA serial dilution: an investigation toward accurate broadband plasmonic signature 22

学校读者我要写书评

暂无评论

DNA serial dilution: an investigation toward accurate broadb...

Plasmonics in Biology and Medicine XXII 2025

作者： Dehzangi, Arash Ghasemi, Masih Choudhury, P.K. Department of Electrical and Computer Engineering University of Texas at Dallas RichardsonTX United States Laser and Plasma Research Institute Shahid Beheshti University Daneshju Blvd.Evin Tehran Iran College of Optical Science and Engineering International Research Center for Advanced Photonics Zhejiang University Building 1A 718 East Haizhou Rd. Zhejiang Haining China

ISBN: (纸本)9781510684225

Serial dilution of Balb/c rat DNA in the visible broadband plasmonic spectroscopy was explored as a promising technique to reduce this genomic substance concentration in the nanoscale. The physical DNA of a living creature, after extraction from the resources (such as blood, nail, hair, tissue, and other possible organs) assumes amorphous pile of strands in the microscale or larger. The serial dilution technique was exploited to exponentially decrease the DNA concentration of BALB/c rat - the target living mammal - in the nanoscale, and the diluted DNA was deposited over a thin film of gold (for fixturing and removing moisture). Plasmonic surface waves of different frequencies were found to promote wave propagation more efficiently when lower thickness of unstructured-nanoscale-DNA layer was deposited (over gold thin film). The results manifest that the most diluted DNA sample reveals distinguishable plasmonic conditions. The sensitivity of the configuration was enhanced in the presence of a DNA sample (as compared to the case of non-existing measurand), which was further increased for larger incidence angles. © 2025 SPIE.

关键词： Surface plasmon resonance

Exponential quantum speedup in simulating coupled classical oscillators*

学校读者我要写书评

暂无评论

Exponential quantum speedup in simulating coupled classical ...

Annual IEEE Symposium on Foundations of computer science

作者： Ryan Babbush Dominic W. Berry Robin Kothari Rolando D. Somma Nathan Wiebe Google Quantum AI Venice CA United States School of Mathematical and Physical Sciences Macquarie University Sydney NSW Australia Department of Computer Science University of Toronto Toronto ON Canada Pacific Northwest National Laboratory Richland WA United States Canadian Institute for Advanced Research Toronto ON Canada

We study the problem of simulating the time evolution of a system of 2 n classical coupled oscillators (e.g., 2 n balls connected by springs) on a quantum computer. We map Newton’s equation for harmonic potentials to Schrödinger’s equation, such that the amplitudes of an $\mathcal{O}(n)$-qubit quantum state encode the momenta and displacements of the 2 n classical oscillators. Given oracle access to the masses and spring constants, we describe a quantum algorithm with query and time complexity poly (n) that solves this problem when certain parameters are polynomially bounded and the initial state is easy to prepare. As an example application, we apply our quantum algorithm to efficiently estimate the normalized kinetic energy of an oscillator at any time. We then show that any classical algorithm solving the same problem must make $2^{\Omega(n)}$ queries to the oracle and we also show that when the oracles are instantiated by poly (n)-size circuits, the problem is BQP-complete. Thus, our approach solves a potentially practical application with an exponential speedup over classical computers.

关键词：