检索结果-内蒙古大学图书馆

A Multi-Modal Behavior Quantitative Analysis Model for Autism Early Screening

学校读者我要写书评

暂无评论

A Multi-Modal Behavior Quantitative Analysis Model for Autis...

2023 IEEE International Conference on Systems, Man, and Cybernetics, SMC 2023

作者： Lei, Jiayi Zhang, E. She, Yingying Wang, Xin Liao, Yuhan Hu, Bin Wu, Hang Yang, Minqiang Tian, Jiajia Wang, Yong School of Informatics Xiamen University China University of Kansas Medical Center Department of Occupational Therapy Education United States School of Film National Institute for Data Science in Health and Medicine Xiamen University China School of Computer Science and Technology Beijing Institute of Technology China Together Education Institute China School of Information Science and Engineering Lanzhou University China Fujian Medical University Union Hospital Pediatric Department China

ISBN: (纸本)9798350337020

Human-computer Interaction (HCI) and Machine Learning (ML) technologies have potential for the behavioral screening of autistic children but how to design a tool and analyse behavior reliably is challenging. Based on psychophysiological computation, this paper proposes an interactive behavior perception analytical model for autism screening. We presented the multi-scenario reactive behavior paradigms that designed based on the atypical characteristics of autistic children. We recorded the eye movement data and facial data of 91 participants, and performed multi-modal feature extraction, used machine learning to train classification model. We conducted comparative experiments, and the experimental results verified the advantages of multi-scenario paradigms and multi-modal feature groups, which indicates that our analysis methods and screening models are effective and reliable and have real research significance. © 2023 IEEE.

关键词：

Newsgroup Topic Extraction using Probabilistic Inverse Cluster Frequency Term-Cluster Weighting and Growing Neural Gas Clustering

学校读者我要写书评

暂无评论

IAENG International Journal of Computer Science

IAENG International Journal of computer Science 2021年第1期48卷 8页

作者： Adinugroho, Sigit Rahman, Muh Arif Syauqy, Dahnial Department of Informatics Faculty of Computer Science Brawijaya University Malang Indonesia Department of Computer Engineering Faculty of Computer Science Brawijaya University Malang Indonesia

Effectiveness of a topic extraction method is depends on the capability to extract information from a large amount of data. This paper proposes a new solution for selecting a set of topics from text documents as well as a new approach to set the weight of a term in clusters of documents. The process of revealing topics from multiple documents starts with a preprocessing step, which aims to omit the unnecessary portion of the text. After that, the weights of terms in documents are calculated based on the term frequency-inverse document frequency method. Then, feature transformation based on singular value decomposition is employed to build weight for clustering. The clustering process is conducted using the Growing Neural Gas method. Finally, to determine the weights of terms in clusters as a way of selecting topics from clusters, the proposed probabilistic inverse cluster frequency term-cluster method is applied. Experiments show that the framework attains satisfactory results indicated by the average accuracy of 0.8606, 0.7406, 0.4039, and 0.6647 for topics obtained from Binary2, Multi5, Multi7, and Multi10 categories of 20Newsgroup dataset. © 2021, IAENG International Journal of computer Science. All Rights Reserved.

关键词： Singular value decomposition

Enhancing Graph Representation Learning with Localized Topological Features

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Yan, Zuoyu Zhao, Qi Ye, Ze Ma, Tengfei Gao, Liangcai Tang, Zhi Wang, Yusu Chen, Chao Wangxuan Institute of Computer Technology Peking University China Computer Science and Engineering Department University of California San Diego United States Department of Biomedical Informatics Stony Brook University United States Halıcıoğlu Data Science Institute University of California San Diego United States Weill Cornell Medicine Cornell University United States

Representation learning on graphs is a fundamental problem that can be crucial in various tasks. Graph neural networks, the dominant approach for graph representation learning, are limited in their representation power. Therefore, it can be beneficial to explicitly extract and incorporate high-order topological and geometric information into these models. In this paper, we propose a principled approach to extract the rich connectivity information of graphs based on the theory of persistent homology. Our method utilizes the topological features to enhance the representation learning of graph neural networks and achieve state-of-the-art performance on various node classification and link prediction benchmarks. We also explore the option of end-to-end learning of the topological features, i.e., treating topological computation as a differentiable operator during learning. Our theoretical analysis and empirical study provide insights and potential guidelines for employing topological features in graph learning tasks. Copyright © 2025, The Authors. All rights reserved.

关键词： Contrastive Learning

IDA: Informed Domain Adaptive Semantic Segmentation

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Chen, Zheng Ding, Zhengming Gregory, Jason M. Liu, Lantao Luddy School of Informatics Computing and Engineering Indiana University BloomingtonIN47408 United States The Department of Computer Science Tulane University United States DEVCOM Army Research Laboratory United States

Mixup-based data augmentation has been validated to be a critical stage in the self-training framework for unsupervised domain adaptive semantic segmentation (UDA-SS), which aims to transfer knowledge from a well-annotated (source) domain to an unlabeled (target) domain. Existing self-training methods usually adopt the popular region-based mixup techniques with a random sampling strategy, which unfortunately ignores the dynamic evolution of different semantics across various domains as training proceeds. To improve the UDA-SS performance, we propose an Informed Domain Adaptation (IDA) model, a self-training framework that mixes the data based on class-level segmentation performance, which aims to emphasize small-region semantics during mixup. In our IDA model, the class-level performance is tracked by an expected confidence score (ECS). We then use a dynamic schedule to determine the mixing ratio for data in different domains. Extensive experimental results reveal that our proposed method is able to outperform the state-of-the-art UDA-SS method by a margin of 1.1 mIoU in the adaptation of GTA-V to Cityscapes and of 0.9 mIoU in the adaptation of SYNTHIA to Cityscapes. Code link: https://***/ArlenCHEN/*** Copyright © 2023, The Authors. All rights reserved.

关键词： Semantics

Improved method of segmentation of images from space-based optoelectronic observation systems based on Otsu's algorithm with global and adaptive thresholds 4

学校读者我要写书评

暂无评论

Improved method of segmentation of images from space-based o...

4th IEEE KhPI Week on Advanced Technology, KhPI Week 2023

作者： Khudov, Hennadii Kalimulin, Temir Khudov, Rostyslav Butko, Ihor Burtseva, Viktoriia Burtsev, Valeriy Department of Radar Troops Tactic Ivan Kozhedub Kharkiv National Air Force University Kharkiv Ukraine Department of Theoretical and Applied Informatics V.N.Karazin Kharkiv National University Kharkiv Ukraine Department of Computer Sciences and Software Engineering Academician Y. Bugay International Scientific and Technical University Kyiv Ukraine Scientific Research Department of Military Standards Metrology Center of Military Standards Kharkiv Ukraine Department of Control Systems of Antiaircraft Missile Troops Ivan Kozhedub Kharkiv National Air Force University Kharkiv Ukraine

ISBN: (纸本)9798350395532

An analysis of optoelectronic image, received from space-based observation system has been carried out. It has been found out, that this image has a complex structure and a unimodal intensity histogram. Otsu method is chosen as to segment the image. This method is easy and relatively quick, when talking of its realization. Methods of global and adaptive thresholding have been researched to perform image segmentation. Results of experimental research of segmentation of optoelectronic image from space-based observation system with use of Otsu method are developed. Ways for future research in this area are defined. © 2023 IEEE.

关键词： Image segmentation

Explainability-Driven Leaf Disease Classification using Adversarial Training and Knowledge Distillation

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Echim, Sebastian-Vasile Tăiatu, Iulian-Marius Cercel, Dumitru-Clementin Pop, Florin Computer Science and Engineering Department National University of Science and Technology POLITEHNICA Bucharest Romania National Institute for Research and Development in Informatics - ICI Bucharest Romania Academy of Romanian Scientists Bucharest Romania

This work focuses on plant leaf disease classification and explores three crucial aspects: adversarial training, model explainability, and model compression. The models’ robustness against adversarial attacks is enhanced through adversarial training, ensuring accurate classification even in the presence of threats. Leveraging explainability techniques, we gain insights into the model’s decision-making process, improving trust and transparency. Additionally, we explore model compression techniques to optimize computational efficiency while maintaining classification performance. Through our experiments, we determine that on a benchmark dataset, the robustness can be the price of the classification accuracy with performance reductions of 3%-20% for regular tests and gains of 50%-70% for adversarial attack tests. We also demonstrate that a student model can be 15-25 times more computationally efficient for a slight performance reduction, distilling the knowledge of more complex models. © 2023, CC BY-NC-SA.

关键词： Distillation

Synergizing Airborne Non-Terrestrial Networks and Reconfigurable Intelligent Surfaces-Aided 6G IoT

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Jamshed, Muhammad Ali Kaushik, Aryan Toka, Mesut Shin, Wonjae Shakir, Muhammad Zeeshan Dash, Soumya P. Dardari, Davide The College of Science and Engineering University of Glasgow United Kingdom The School of Engineering & Informatics University of Sussex United Kingdom The Department of Electrical and Computer Engineering Ajou University Korea Republic of The School of Electrical Engineering Korea University Korea Republic of The Department of Electrical and Computer Engineering Princeton University United States The School of Computing Engineering and Physical Sciences University of the West of Scotland United Kingdom The School of Electrical Sciences Indian Institute of Technology Bhubaneswar India The Department of Electrical Electronic and Information Engineering University of Bologna Italy

On the one hand, Reconfigurable Intelligent Surfaces (RISs) emerge as a promising solution to meet the demand for higher data rates, improved coverage, and efficient spectrum utilization. On the other hand, Non-Terrestrial Networks (NTNs) offer unprecedented possibilities for global connectivity. Moreover, the NTN can also support the upsurge in the number of Internet of Things (IoT) devices by providing reliable and ubiquitous connectivity. Although NTNs have shown promising results, there are several challenges associated with their usage, such as signal propagation delays, interference, security, etc. In this article, we have discussed the possibilities of integrating RIS with an NTN platform to overcome the issues associated with NTN. Furthermore, through experimental validation, we have demonstrated that the RIS-assisted NTN can play a pivotal role in improving the performance of the entire communication system. © 2023, CC BY.

关键词： Internet of things

Near-Field Localization With Extremely Large Scale Antenna Arrays in the Presence of Mutual Coupling

学校读者我要写书评

暂无评论

Near-Field Localization With Extremely Large Scale Antenna A...

European Conference on Antennas and Propagation, EuCAP

作者： Zohreh Ebadi Amir Masoud Molaei George C. Alexandropoulos Muhammad Ali Babar Abbasi Simon Cotton Anvar Tukmanov Okan Yurduseven Centre for Wireless Innovation EEECS Queen's University Belfast Belfast UK Department of Informatics and Telecommunications National and Kapodistrian University of Athens Athens Greece Department of Electrical and Computer Engineering University of Illinois Chicago IL USA BT Labs Adastral Park BT Group UK

ISBN: (数字)9788831299107

ISBN: (纸本)9798350366327

Extremely large scale antenna arrays (ELAAs) will be a key technology in sixth-generation (6G) wireless communication systems. Due to the large near-field region of ELAAs, sources can easily fall within this area. Hence, a near-field localization algorithm that accounts for the electromagnetic effects of ELAA elements, such as mutual coupling (MC), is essential for proper 6G wireless communication systems design. The three-dimensional (3D) multiple signal classification (MUSIC) algorithm can be used to estimate the near-field source location in the presence of MC. However, this method suffers from a high computational load due to the requirement to perform 3D-based search steps. Recently, to address this issue, an iterative method based on an oblique projection operator (IMOP) has been presented. However, IMOP's computational load can still be high, particularly for ELAA-related scenarios. In this paper, we address this issue by using one sub-array of ELAA which holds far-field assumptions to estimate the direction of arrival (DOA). The DOA estimation relies on an one-dimensional (1D) MUSIC-based algorithm, drastically reducing the computational load. Then, the MC coefficients are calculated using the estimated DOAs. Finally, another 1D MUSIC step is used to estimate the ranges corresponding to the estimated DOA and MC coefficients. Our representative simulation results verify the superiority of the proposed near-field source localization in the presence of MC in terms of computational load, while offering comparable accuracy with the IMOP method.

关键词： Location awareness Wireless communication Mutual coupling Direction-of-arrival estimation Accuracy Simulation Estimation Multiple signal classification Classification algorithms Antenna arrays

Defect Recognition based on Radon Transform in Pentacene organic thin-film Archimedean interdigitated spirals transistor

学校读者我要写书评

暂无评论

Defect Recognition based on Radon Transform in Pentacene org...

International Conference on Clean Electrical Power (ICCEP)

作者： Grazia Lo Sciuto Sivan Linde Rafi Shikler Paweł Kowol Salvo Coco Giacomo Capizzi Department of Mechatronics Silesian University of Technology Gliwice Poland Department of Electrical Electronics and Informatics Engineering University of Catania Catania Italy Department of Electrical and Computer Engineering School of Electrical and Computer Engineering Optoelectronic Organic Semiconductor Devices Laboratory Ben-Gurion University of the Negev Beer-Sheva Israel Ilse Katz Institute for Nanoscale Science and Technology Ben-Gurion University of the Negev Beer-Sheva Israel

In this paper the problem of automatic defects detection in thin film transistor is considered in order to increase the efficiency of electronic devices. An effective approach based on defect pattern recognition in organic transistor is developed using the Radon transform. The adopted design and technology in Pentacene organic thin-film transistor exhibits two interdigitated Archimedean spirals pattern. In the conducted experiments, the images acquisition is performed with a microscope camera to capture the various defects of surface morphology on the top of the examined transistors. The defects detection is obtained by using the Radon transform on each channel (R,G,B) of these images.

关键词：