检索结果-内蒙古大学图书馆

arXiv 2024年

作者： Mattamala, Matias Frey, Jonas Libera, Piotr Chebrolu, Nived Martius, Georg Cadena, Cesar Hutter, Marco Fallon, Maurice Dynamic Robot Systems Group University of Oxford 23 Banbury Road Oxford OxfordshireOX2 6NN United Kingdom Robotic Systems Lab ETH Zurich Leonhardstrasse 21 Zurich Zurich8092 Switzerland Autonomous Learning Group Max Planck Institute for Intelligent Systems Max-Planck-Ring 4 Tübingen Baden-Württemberg72076 Germany Computer Science Department University of Tüüingen Maria-von-Linden-Strasse 6 Tübingen Baden-Württemberg72076 Germany

Natural environments such as forests and grasslands are challenging for robotic navigation because of the false perception of rigid obstacles from high grass, twigs, or bushes. In this work, we present Wild Visual Navigation (WVN), an online self-supervised learning system for visual traversability estimation. The system is able to continuously adapt from a short human demonstration in the field, only using onboard sensing and computing. One of the key ideas to achieve this is the use of high-dimensional features from pre-trained self-supervised models, which implicitly encode semantic information that massively simplifies the learning task. Further, the development of an online scheme for supervision generator enables concurrent training and inference of the learned model in the wild. We demonstrate our approach through diverse real-world deployments in forests, parks, and grasslands. Our system is able to bootstrap the traversable terrain segmentation in less than 5 min of in-field training time, enabling the robot to navigate in complex, previously unseen outdoor terrains. © 2024, CC BY.

关键词： Parks

来源：评论

学校读者我要写书评

暂无评论

Federated Learning Robust to Byzantine Attacks: Achieving Zero Optimality Gap

arXiv

引用

arXiv 2023年

作者： Zuo, Shiyuan Fan, Rongfei Hu, Han Zhang, Ning Gong, Shimin The School of Cyberspace Science and Technology Beijing Institute of Technology Beijing100081 China The School of Information and Electronics Beijing Institute of Technology Beijing100081 China The Department of Electrical and Computer Engineering University of Windsor WindsorONN9B 3P4 Canada The School of Intelligent Systems Engineering Sun Yat-sen University Shenzhen518055 China

In this paper, we propose a robust aggregation method for federated learning (FL) that can effectively tackle malicious Byzantine attacks. At each user, model parameter is firstly updated by multiple steps, which is adjustable over iterations, and then pushed to the aggregation center directly. This decreases the number of interactions between the aggregation center and users, allows each user to set training parameter in a flexible way, and reduces computation burden compared with existing works that need to combine multiple historical model parameters. At the aggregation center, geometric median is leveraged to combine the received model parameters from each user. Rigorous proof shows that zero optimality gap is achieved by our proposed method with linear convergence, as long as the fraction of Byzantine attackers is below half. Numerical results verify the effectiveness of our proposed method. Copyright © 2023, The Authors. All rights reserved.

关键词： Machine learning

来源：评论

学校读者我要写书评

暂无评论

Look closer to segment better: Boundary patch refinement for instance segmentation

arXiv

引用

arXiv 2021年

作者： Tang, Chufeng Chen, Hang Li, Xiao Li, Jianmin Zhang, Zhaoxiang Hu, Xiaolin State Key Laboratory of Intelligent Technology and Systems THU-Bosch JCML Center BNRist Institute for AI Department of Computer Science and Technology Tsinghua University China Institute of Automation CAS University of Chinese Academy of Sciences Centre for Artificial Intelligence and Robotics HKISI CAS China

Tremendous efforts have been made on instance segmentation but the mask quality is still not satisfactory. The boundaries of predicted instance masks are usually imprecise due to the low spatial resolution of feature maps and the imbalance problem caused by the extremely low proportion of boundary pixels. To address these issues, we propose a conceptually simple yet effective post-processing refinement framework to improve the boundary quality based on the results of any instance segmentation model, termed BPR. Following the idea of looking closer to segment boundaries better, we extract and refine a series of small boundary patches along the predicted instance boundaries. The refinement is accomplished by a boundary patch refinement network at higher resolution. The proposed BPR framework yields significant improvements over the Mask R-CNN baseline on Cityscapes benchmark, especially on the boundary-aware metrics. Moreover, by applying the BPR framework to the "PolyTransform + SegFix" baseline, we reached 1st place on the Cityscapes leaderboard. Code is available at https://***/tinyalpha/BPR. Copyright © 2021, The Authors. All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Ontology-based data interestingness: A state-of-the-art review

Natural Language Processing Journal

引用

Natural Language Processing Journal 2023年 4卷

作者： Abhilash C.B. Kavi Mahesh Department of Computer Science and Engineering Indian Institute of Information Technology Dharwad India Department of Data Science and Intelligent Systems Indian Institute of Information Technology Dharwad India

In recent years, there has been significant growth in the use of ontology-based methods to enhance data interestingness. These methods play a crucial role in knowledge management by providing explicit, shareable, and reusable knowledge descriptions. However, limited research has been conducted to explore the impact and practical benefits of employing ontologies to increase data interestingness. A comprehensive analysis of the current state-of-the-art approaches in ontology-based data interestingness is essential to identify future advancements and directions in this field. The main objective of this study is to present a clear roadmap showcasing how ontologies contribute to data interestingness and to evaluate their effectiveness in the existing literature. The findings of this research offer an integrated and comprehensive understanding of the following aspects: (1) interestingness metrics in data mining, (2) association rule mining to identify interesting patterns, (3) ontology-based methods to improve data interestingness, and (4) techniques that use ontologies to derive interestingness in data. The paper concludes with a summary of ontology-based methods for data interestingness and outlines potential avenues for future research.

关键词： Ontology Data interestingness Knowledge management Semantics Association rule mining Data mining Interestingness metrics Ontology-based methods

来源：评论

学校读者我要写书评

暂无评论

High-Precision 3D Printing by Deploying Expandable Microspheres

SSRN

引用

SSRN 2024年

作者： Xu, Yang Wang, Fucheng Zhang, Pei Lu, Chengbang Liu, Weixuan Guo, Yueying Ni, Zhipeng Wang, Jinqing Lu, Baoyang Lou, Qin Liang, Xiangyu Liu, Ji Department of Mechanical and Energy Engineering Southern University of Science and Technology Shenzhen518055 China Department of Metrology and Measurement Engineering China Jiliang University Hangzhou310018 China Agricultural Genomics Institute at Shenzhen Chinese Academy of Agricultural Sciences Shenzhen518120 China Institute of Bast Fiber Crops Center of Southern Economic Crops Chinese Academy of Agricultural Sciences Changsha410205 China Department of Mechanics and Aerospace Engineering Southern University of Science and Technology Shenzhen518055 China Yiwu Saiens Polymer Materials Co. Ltd Jinhua322099 China Jiangxi Key Lab of Flexible Electronics Flexible Electronics Innovation Institute Jiangxi Science & Technology Normal University Nanchang330013 China Shenzhen Key Laboratory of Intelligent Robotics and Flexible Manufacturing Systems Southern University of Science and Technology Shenzhen518055 China

Stereolithography-based 3D printing (3DP) technology is widely employed in various industries, including manufacturing, healthcare, energy, biomedical, art and other fields. However, the precision issues, such as dimensional shrinkage and structural warping, significantly hinder its wide applications. In this study, we present a straightforward, efficient yet general strategy to engineer the photosensitive resins with thermally expandable microspheres, thus enabling significant improvement in structural fidelity of the 3DP constructs. Our findings demonstrate that addition of expandable microspheres substantially mitigates the volume shrinkage below 3.98%, in comparison to over 10% for commercially-available photosensitive resins. An exceptionally high precision of 0.035% in dimensional deviation could be achieved, in comparison to over 0.1% for commercial one. Furthermore, due to the low filling ratio, the improvement in 3DP precision does not sacrifice the mechanical properties, thus it does not affect the applications where those photosensitive resins are originally targeting. Our method represents an effective strategy to improve the 3D printing resolution of photosensitive resins, thus opening new directions for high precision 3D printing technology. © 2024, The Authors. All rights reserved.

关键词： Shrinkage

来源：评论

学校读者我要写书评

暂无评论

High-Precision Stereo Matching with a Drone Equipped With Rtk-Gnss and a Single Camera for Plant Height Measurements

SSRN

引用

SSRN 2022年

作者： Matsuura, Yuta Heming, Zhang Nakao, Kousuke Qiong, Chang Firmansyah, Iman Kawai, Shin Yamaguchi, Yoshiki Hayashi, Hisayoshi Maruyama, Tsutomu Nobuhara, Hajime The Department of Intelligent and Mechanical Interaction Systems Graduate School of Science and Technology University of Tsukuba 1-1-1 Tennoudai Ibaraki Tsukuba305-8573 Japan College of Engineering Systems School of Science and Engineering University of Tsukuba Japan The Department of Computer Science Graduate School of Science and Technology University of Tsukuba Japan The Faculty of Life and Environmental Sciences University of Tsukuba Japan The School of Computing Tokyo Institute of Technology Japan

We propose a high-precision plant height measurement system based on large-scale stereo images captured by a drone. In our system, the drone collects images of a field using a monocular camera and obtains the position data by a real time kinematic (RTK)-global navigation satellite system (GNSS). The heights of plants in the field are measured by stereo matching. By employing a single monocular camera rather than a binocular camera, we can obtain a sufficient baseline length for accurate height measurements, but the images must be precisely calibrated at run time. To this end, we propose a calibration algorithm with a conventional stereo matching algorithm. The proposed method requires only a pair of images and their position data (RTK-GNSS), which are more suitable for edge computing on drones than three-dimensional (3D) reconstructions by methods such as Structure from Motion. To show the effectiveness of the proposed method, we measured the plant height from photographs captured by the drone at different altitudes. Our method achieves a higher height resolution (1.6 cm) than the 3D reconstruction-based methods Pix4Dmapper and Metashape, with error-rate reductions of 62.2% and 69.4%, respectively, in the 10-20 m altitude range. When the altitude is lowered to 4.1 m , the depth resolution is improved to 1.6 mm with error reductions of 44.4% and 63.0% from those of Pix4Dmapper and Metashape, respectively. On a GTX 1080 TI graphics processing unit, stereo matching of 5472 × 3468-pixel images consumed 88 ms. This execution time is sufficiently fast for real-time measurements. Because our system uses the position data obtained by RTK-GNSS, it achieves centimeter-level accuracy during the drone flight. After the flight, the accuracy can be improved to millimeter level by comparing the results of the stereo matching with an image containing a reference plant of preobtained known height © 2022, The Authors. All rights reserved.

关键词： Global positioning system

来源：评论

学校读者我要写书评

暂无评论

No One Left Behind: Real-World Federated Class-Incremental Learning

arXiv

引用

arXiv 2023年

作者： Dong, Jiahua Li, Hongliu Cong, Yang Sun, Gan Zhang, Yulun Van Gool, Luc The State Key Laboratory of Robotics Shenyang Institute of Automation Chinese Academy of Sciences Shenyang110016 China The Institutes for Robotics and Intelligent Manufacturing Chinese Academy of Sciences Shenyang110169 China The University of Chinese Academy of Sciences Beijing100049 China The Department of Civil and Environmental Engineering Hong Kong Polytechnic University Hong Kong The College of Automation Science and Engineering South China University of Technology Guangzhou510640 China The Computer Vision Lab ETH Zürich Zürich8092 Switzerland

Federated learning (FL) is a hot collaborative training framework via aggregating model parameters of decentralized local clients. However, most FL methods unreasonably assume data categories of FL framework are known and fixed in advance. Moreover, some new local clients that collect novel categories unseen by other clients may be introduced to FL training irregularly. These issues render global model to undergo catastrophic forgetting on old categories, when local clients receive new categories consecutively under limited memory of storing old categories. To tackle the above issues, we propose a novel Local-Global Anti-forgetting (LGA) model. It ensures no local clients are left behind as they learn new classes continually, by addressing local and global catastrophic forgetting. Specifically, considering tackling class imbalance of local client to surmount local forgetting, we develop a category-balanced gradient-adaptive compensation loss and a category gradient-induced semantic distillation loss. They can balance heterogeneous forgetting speeds of hard-to-forget and easy-to-forget old categories, while ensure consistent class-relations within different tasks. Moreover, a proxy server is designed to tackle global forgetting caused by Non-IID class imbalance between different clients. It augments perturbed prototype images of new categories collected from local clients via self-supervised prototype augmentation, thus improving robustness to choose the best old global model for local-side semantic distillation loss. Experiments on representative datasets verify superior performance of our model against comparison methods. The code is available at https://***/JiahuaDong/LGA. 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

Enhancement of galena-potassium ethyl xanthate flotation system by low power ultrasound

引用

Transactions of Nonferrous Metals Society of China 2020年第4期30卷 1102-1110页

作者： Can GUNGOREN Yasin BAKTARHAN Ismail DEMIR Safak Gokhan OZKAN Department of Mining Engineering Engineering FacultyIstanbul University-CerrahpasaBuyukcekmeceIstanbul 34500Turkey Department of Robotics and Intelligent Systems The Institute of the Graduate Studies in Science and EngineeringTurkish-German UniversityBeykozIstanbul 34820Turkey

The objective of this study is to investigate the improvement possibilities of the floatability of galena with ultrasonic application in the presence of potassium ethyl xanthate(KEX). For this purpose, micro-flotation experiments were carried out in addition to surface chemistry studies including zeta potential, contact angle, and bubble-particle attachment time measurements at various ultrasonic power levels and conditioning time. The results showed that, the maximum micro-flotation recovery of 77.5% was obtained with 30 W ultrasound power and 2 min conditioning time. In addition, more negative zeta potential values were obtained with ultrasound as well as higher contact angle and lower bubble-particle attachment time, which indicated the increased hydrophobicity of galena with ultrasound.

关键词： galena potassium ethyl xanthate flotation ultrasound bubble-particle attachment time

来源：评论

学校读者我要写书评

暂无评论

A Data logger for educational purposes of a laboratory chemical reactor: an IoT approach

A Data logger for educational purposes of a laboratory chemi...

引用

Electrical, computer, Communications and Mechatronics Engineering (ICECCME), International Conference on

作者： José Lima Thadeu Brito Olga Ferreira Maria J. Afonso Vítor H. Pinto José A. Carvalho Paulo Costa Research Centre in Digitalization and Intelligent Robotics (CeDRI) Instituto Politécnico de Bragança Bragança Portugal Laboratório Para a Sustentabilidade e Tecnologia em Regiões de Montanha (SusTEC) Instituto Politécnico de Bragança Bragança Portugal Centre for Robotic in Industry and Intelligent Systems Institute for Systems and Computer Engineering Technology and Science (INESC-TEC) Porto Portugal CeDRI SusTEC INESC-TEC Faculty of Engineering of University of Porto (FEUP) Porto Portugal SusTEC ESTiG Instituto Politécnico de Bragança Bragança Portugal ARISE & ECE Dept. Research Center for Systems and Technologies (SYSTEC) FEUP CeDRI SusTEC INESC-TEC FEUP

This paper presents the development of an acquisition system and data logger from an existing set of three continuous stirred-tank reactors in series. The reactors are currently used in chemical engineering educational laboratories to perform kinetic and tracer experiments. In this sense, to accomplish the store data process, the volumetric flow rate and the concentration of tracer, reactants and/or products of the reaction must be acquired as a function of time. In the original experimental setup, only the signal conditioning system was operational, while the acquisition, visualization, and control systems were obsolete and damaged. Thus, a new system composed of an interface and real-time acquisition data is proposed alongside preserving the main reactor structure. A graphical user interface and the automation of the various actuators were developed based on worldwide usage and low cost, respectively. This system, based on a common 8-bit microcontroller and an application developed in Lazarus, allows the storage of the acquired data in a time-series database. In this way, students can analyze the results later or in real time. Moreover, remote access allows controlling the reactor and getting data by the Internet of Things (IoT) resources. Additionally, the proposed system using IoT allows data to be shared with the community as a dataset.

关键词：

来源：评论

学校读者我要写书评

暂无评论

CARDIOCARE platform: A beyond the state of the art approach for the management of elderly multimorbid patients with breast cancer therapy induced cardiac toxicity

CARDIOCARE platform: A beyond the state of the art approach ...

引用

2023 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2023

作者： Tsiouris, Kostas M. Kalliatakis, Grigoris Mazzocco, Ketti Seruga, Bostjan Marias, Kostas Karanasiou, Georgia Antoniades, Athos Papakonstantinou, Andri Conti, Constanza Tsiknakis, Manolis Sfakianakis, Stelios Lakkas, Lampros Filippatos, Gerasimos Bucur, Anca Fotiadis, Dimitrios I. Manikis, George Mauri, Davide Constantinidou, Anastasia Pacella, Elsa University of Ioannina Unit of Medical Technology and Intelligent Information Systems Ioannina Greece Foundation for Research and Technology - Hellas Computational Bio-Medicine Laboratory Institute of Computer Science Heraklion Greece European Institute of Oncology Cardiology Division Milan Italy Institute of Oncology Ljubljana Department of Medical Oncology Ljubljana Slovenia Stremble Ventures Ltd Limassol Cyprus Karolinska University Hospital Endocrine Tumors and Sarcoma Theme Cancer Dept. for Breast Cancer Sweden Istituto di Management Sanitario S.r.l Milano Italy Foundation for Research and Technology - Hellas Institute of Computer Science Heraklion Greece University of Ioannina 2nd Dept. of Cardiology School of Medicine Ioannina Greece National and Kapodistrian University of Athens Athens Greece Attikon University Hospital School of Medicine Dept. of Cardiology Athens Greece Philips Research Europe Eindhoven Netherlands University of Ioannina Dept. of Medical Oncology Ioannina Greece Bank of Cyprus Oncology Centre Nicosia Cyprus European Society of Cardiology Brussels Belgium

ISBN: (纸本)9798350337488

Breast cancer (BC) is the most common cancer in women in Europe and worldwide, with a high prevalence in middle-aged and older women. The last years, the evolution in the existing treatment approaches have contributed to improved clinical outcomes and survival rates. Nevertheless, BC therapy-related cardiotoxicity, poses a severe impact in the short- and long-term Quality of Life (QoL) and associated survival of the BC patients. This study demonstrates how the CARDIOCARE platform and the developed risk stratification models provides healthcare professionals with a valuable tool for effectively managing BC patients, preventing treatment induced cardiotoxicity and improving their QoL. This is accomplished through the integration of multi-source patient-specific data from patient-oriented mobile applications and wearable sensors, and by the employment of beyond the state-of-the-art data mining and machine learning approaches. © 2023 IEEE.

关键词： Breast cancer cardiotoxicity machine learning mobile application risk stratification models

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：