检索结果-内蒙古大学图书馆

RefineMask: Towards high-quality instance segmentation with fine-grained features

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Zhang, Gang Lu, Xin Tan, Jingru Li, Jianmin Zhang, Zhaoxiang Li, Quanquan Hu, Xiaolin State Key Laboratory of Intelligent Technology and Systems Institute for AI BNRist Department of Computer Science and Technology Tsinghua University SenseTime Research Tongji University Institute of Automation CAS & UCAS

The two-stage methods for instance segmentation, e.g. Mask R-CNN, have achieved excellent performance recently. However, the segmented masks are still very coarse due to the downsampling operations in both the feature pyramid and the instance-wise pooling process, especially for large objects. In this work, we propose a new method called RefineMask for high-quality instance segmentation of objects and scenes, which incorporates fine-grained features during the instance-wise segmenting process in a multi-stage manner. Through fusing more detailed information stage by stage, RefineMask is able to refine high-quality masks consistently. RefineMask succeeds in segmenting hard cases such as bent parts of objects that are over-smoothed by most previous methods and outputs accurate boundaries. Without bells and whistles, RefineMask yields significant gains of 2.6, 3.4, 3.8 AP over Mask R-CNN on COCO, LVIS, and Cityscapes benchmarks respectively at a small amount of additional computational cost. Furthermore, our single-model result outperforms the winner of the LVIS Challenge 2020 by 1.3 points on the LVIS test-dev set and establishes a new state-of-the-art. Code will be available at https://***/zhanggang001/RefineMask. Copyright © 2021, The Authors. All rights reserved.

关键词：

Evaluation of Robustness Metrics for Defense of Machine Learning systems

学校读者我要写书评

暂无评论

Evaluation of Robustness Metrics for Defense of Machine Lear...

International Conference on Military Communications and Information systems (ICMCIS)

作者： J. DeMarchi R. Rijken J. Melrose B. Madahar G. Fumera F. Roli E. Ledda M. Aktaş F. Kurth P. Baggenstoss B. Pelzer L. Kanestad Collaborative Engineering Systems & Aerospace Systems Information Supremacy Royal Netherlands Aerospace Centre NLR Amsterdam NLD Cyber & Information Systems Division Defence Science and Technology Laboratory Portondown GBR Department of Electrical and Electronic Engineering University of Cagliari Cagliari ITA Department of Informatics Bioengineering Robotics and Systems Engineering University of Genoa Genoa ITA Department of Computer Control and Management Engineering Sapienza University of Rome Rome ITA Defence Systems Technologies Division ASELSAN Ankara TUR Fraunhofer Institute for Communication Information Processing and Ergonomics Bonn DEU Swedish Defence Research Agency Cyber Defence and C2 Technology Division Stockholm SWE

In this paper we explore some of the potential applications of robustness criteria for machine learning (ML) systems by way of tangible “demonstrator” scenarios. In each demonstrator, ML robustness metrics are applied to real-world scenarios with military relevance, indicating how they might be used to help detect and handle possible adversarial attacks on ML systems. We conclude by sketching promising future avenues of research in order to: (1) help establish useful verification methodologies to facilitate ML robustness compliance assessment; (2) support development of ML accountability mechanisms; and (3) reliably detect, repel, and mitigate adversarial attack.

关键词：

Adversarial Texture for Fooling Person Detectors in the Physical World

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Hu, Zhanhao Huang, Siyuan Zhu, Xiaopei Sun, Fuchun Zhang, Bo Hu, Xiaolin Department of Computer Science and Technology Institute for Artificial Intelligence State Key Laboratory of Intelligent Technology and Systems BNRist Tsinghua University Beijing China School of Integrated Circuits Tsinghua University Beijing China IDG/McGovern Institute for Brain Research Tsinghua University Beijing China Beijing China

Nowadays, cameras equipped with AI systems can capture and analyze images to detect people automatically. However, the AI system can make mistakes when receiving deliberately designed patterns in the real world, i.e., physical adversarial examples. Prior works have shown that it is possible to print adversarial patches on clothes to evade DNN-based person detectors. However, these adversarial examples could have catastrophic drops in the attack success rate when the viewing angle (i.e., the camera's angle towards the object) changes. To perform a multi-angle attack, we propose Adversarial Texture (AdvTexture). AdvTexture can cover clothes with arbitrary shapes so that people wearing such clothes can hide from person detectors from different viewing angles. We propose a generative method, named Toroidal-Cropping-based Expandable Generative Attack (TC-EGA), to craft AdvTexture with repetitive structures. We printed several pieces of cloth with AdvTexure and then made T-shirts, skirts, and dresses in the physical world. Experiments showed that these clothes could fool person detectors in the physical world. Copyright © 2022, The Authors. All rights reserved.

关键词： Textures

学校读者我要写书评

暂无评论

GUEST EDITORIAL

China Communications 2021年第12期 4-7页

作者： Yingchang Liang Yue Gao Liang Xiao Yulong Zou Guoru Ding the University of Electronic Science and Technology of China the Center for Intelligent Networking and Communications the Artificial Intelligence Research Institute the School of Computer Science and Electronic Engineering Institute for Communication Systems University of Surrey Engineering and Physical Sciences Research Council the IEEE Com Soc Wireless Communication Technical Committee the Department of Information and Communication Engineering Xiamen University the NUPT the College of Communications Engineering PLA Army Engineering University

To meet the rapid growth of mobile communications services, the continuously upgraded wireless networks significantly increase the demand for radio spectrum ***, the spectrum supply is limited, thus triggering the spectrum demand-supply ***, dynamic spectrum sharing emerges as a promising paradigm of enhancing spectrum usage, which has been widely recognized and adopted in commercial wireless standards,

关键词：

Optimal Sensor Placement Using Combinations of Hybrid Measurements for Source Localization

学校读者我要写书评

暂无评论

Optimal Sensor Placement Using Combinations of Hybrid Measur...

IEEE National Conference on Radar

作者： Kang Tang Sheng Xu Yuqi Yang He Kong Yongsheng Ma Shenzhen Key Labor. of Intel. Robo. and Flex. Manuf. Sys. Southern Uni. of Sci. and Tech. (SUSTech) Shenzhen China Chinese Academy of Sciences Shenzhen Institute of Advanced Technology Shenzhen China Department of Computer Science University of Nottingham Nottingham UK Shenzhen Key Laboratory of Control Theory and Intelligent Systems SUSTech Shenzhen China

ISBN: (数字)9798350329209

ISBN: (纸本)9798350329216

This paper focuses on static source localization employing different combinations of measurements, including time-difference-of-arrival (TDOA), received-signal-strength (RSS), angle-of-arrival (AOA), and time-of-arrival (TOA) measurements. Since sensor-source geometry significantly impacts localization accuracy, the strategies of optimal sensor placement are proposed systematically using combinations of hybrid measurements. Firstly, the relationship between sensor placement and source estimation accuracy is formulated by a derived Cramér-Rao bound (CRB). Secondly, the A-optimality criterion, i.e., minimizing the trace of the CRB, is selected to calculate the smallest reachable estimation mean-squared-error (MSE) in a unified manner. Thirdly, the optimal sensor placement strategies are developed to achieve the optimal estimation bound. Specifically, the specific constraints of the optimal geometries deduced by specific measurement, i.e., TDOA, AOA, RSS, and TOA, are found and discussed theoretically. Finally, the new findings are verified by simulation studies.

关键词： Location awareness Geometry Sensor placement Time difference of arrival Radar measurements Noise Estimation

A modular rehabilitation lower limb exoskeleton for stroke patients with hemiplegia

学校读者我要写书评

暂无评论

A modular rehabilitation lower limb exoskeleton for stroke p...

2021 IEEE International Conference on Real-Time Computing and robotics, RCAR 2021

作者： Li, Pengbo Wei, Wenhao Bao, Ruoyu He, Bailin Su, Zhilong Wang, Can Wu, Xinyu Shenzhen Institute of Advanced Technology Chinese Academy of Sciences Guangdong Provincial Key Lab of Robotics and Intelligent System Shenzhen518055 China CAS Key Laboratory of Human-Machine Intelligence-Synergy Systems Shenzhen Institute of Advanced Technology University of Chinese Academy of Sciences Department of Mechanical and Automation Engineering The Chinese University of Hong Kong College of Mechanical and Electrical Engineering Shaanxi University of Science and Technology Shaanxi Xi'an710021 China

ISBN: (纸本)9781665436786

A modular rehabilitation exoskeleton robot for hemiplegic patients (SIAT-H) is created in this paper. The design of the exoskeleton is based on the gait analysis of hemiplegic patients and the movement data of healthy people. Through the motion capture experiments at different speeds, the walking style of patients with hemiplegia is understood. A modular exoskeleton robot, including an independent hip joint, knee joint and ankle joint, is designed. The mechanical structure which is optimized based on the physiological characteristics of human body would not cause additional damage to the joints during walking. In order to ensure the reliability of use, the structural strength is also considered. The kinematics relationship is verified by comparison of simulation data and calculated data. In the end, we made an experimental prototype, and the wearing experiment verified the reliability of the mechanism. © 2021 IEEE.

关键词： Exoskeleton (robotics)

Fair Patient Model: Mitigating Bias in the Patient Representation Learned from the Electronic Health Records

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Sivarajkumar, Sonish Huang, Yufei Wang, Yanshan Intelligent Systems Program School of Computing and Information University of Pittsburgh PittsburghPA United States Department of Electrical and Computer Engineering University of Pittsburgh PittsburghPA United States Department of Health Information Management University of Pittsburgh PittsburghPA United States Department of Biomedical Informatics University of Pittsburgh PittsburghPA United States Clinical and Translational Science Institute University of Pittsburgh PittsburghPA United States

Objective: To pre-train fair and unbiased patient representations from Electronic Health Records (EHRs) using a novel weighted loss function that reduces bias and improves fairness in deep representation learning models. Methods: We defined a new loss function, called weighted loss function, in the deep representation learning model to balance the importance of different groups of patients and features. We applied the proposed model, called Fair Patient Model (FPM), to a sample of 34,739 patients from the MIMIC-III dataset and learned patient representations for four clinical outcome prediction tasks. Results: FPM outperformed the baseline models in terms of three fairness metrics: demographic parity, equality of opportunity difference, and equalized odds ratio. FPM also achieved comparable predictive performance with the baselines, with an average accuracy of 0.7912. Feature analysis revealed that FPM captured more information from clinical features than the baselines. Conclusion: FPM is a novel method to pre-train fair and unbiased patient representations from EHR data using a weighted loss function. The learned representations can be used for various downstream tasks in healthcare and can be extended to other domains where bias and fairness are important. Copyright © 2023, The Authors. All rights reserved.

关键词： Deep learning

Heterogeneous virus classification using a functional deep learning model based on transmission electron microscopy images

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Sikder, Niloy Khan, Al-Masrur Bairagi, Anupam Kumar Masud, Mehedi Tiang, Jun Jiat Nahid, Abdullah-Al Radboud University Medical Center Donders Institute for Brain Cognition and Behaviour Nijmegen Netherlands Faculty of Technology and Bionics Rhine-Waal University of Applied Sciences Kleve Germany Department of Intelligent Systems Engineering Indiana University BloomingtonIN United States Computer Science and Engineering Discipline Khulna University Khulna9208 Bangladesh Department of Computer Science College of Computers and Information Technology Taif University P.O. Box 11099 Taif21944 Saudi Arabia Faculty of Engineering Multimedia University Cyberjaya63100 Malaysia Electronics and Communication Engineering Discipline Khulna University Khulna9208 Bangladesh

Viruses are submicroscopic agents that can infect all kinds of lifeforms and use their hosts’ living cells to replicate themselves. Despite having some of the simplest genetic structures among all living beings, viruses are highly adaptable, resilient, and given the right conditions, are capable of causing unforeseen complications in their hosts’ bodies. Due to their multiple transmission pathways, high contagion rate, and lethality, viruses are the biggest biological threat faced by animal and plant species. It is often challenging to promptly detect the presence of a virus in a possible host’s body and accurately determine its type using manual examination techniques;however, it can be done using computer-based automatic diagnosis methods. Most notably, the analysis of Transmission Electron Microscopy (TEM) images has been proven to be quite successful in instant virus identification. Using TEM images collected from a recently published dataset, this article proposes a deep learning-based classification model to identify the type of virus within those images correctly. The methodology of this study includes two coherent image processing techniques to reduce the noise present in the raw microscopy images. Experimental results show that it can differentiate among the 14 types of viruses present in the dataset with a maximum of 97.44% classification accuracy and F1-score, which asserts the effectiveness and reliability of the proposed method. Implementing this scheme will impart a fast and dependable way of virus identification subsidiary to the thorough diagnostic procedures. © 2024, CC BY.

关键词： computer aided diagnosis

Continuous-time fitted value iteration for robust policies

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Lutter, Michael Belousov, Boris Mannor, Shie Fox, Dieter Garg, Animesh Peters, Jan Intelligent Autonomous Systems group Computer Science Department Technical University of Darmstadt Darmstadt Germany Andrew and Erna Viterbi Faculty of Electrical & Computer Engineering Technion – Israel Institute of Technology and a Distinguished Scientist at NVIDIA Tel Aviv Israel Allen School of Computer Science & Engineering University of Washington and Senior Director of Robotics Research NVIDIA Washington Seattle United States Computer Science Department University of Toronto NVIDIA Toronto Canada

Solving the Hamilton-Jacobi-Bellman equation is important in many domains including control, robotics and economics. Especially for continuous control, solving this differential equation and its extension the Hamilton-Jacobi-Isaacs equation, is important as it yields the optimal policy that achieves the maximum reward on a give task. In the case of the Hamilton-Jacobi-Isaacs equation, which includes an adversary controlling the environment and minimizing the reward, the obtained policy is also robust to perturbations of the dynamics. In this paper we propose continuous fitted value iteration (cFVI) and robust fitted value iteration (rFVI). These algorithms leverage the non-linear control-affine dynamics and separable state and action reward of many continuous control problems to derive the optimal policy and optimal adversary in closed form. This analytic expression simplifies the differential equations and enables us to solve for the optimal value function using value iteration for continuous actions and states as well as the adversarial case. Notably, the resulting algorithms do not require discretization of states or actions. We apply the resulting algorithms to the Furuta pendulum and cartpole. We show that both algorithms obtain the optimal policy. The robustness Sim2Real experiments on the physical systems show that the policies successfully achieve the task in the real-world. When changing the masses of the pendulum, we observe that robust value iteration is more robust compared to deep reinforcement learning algorithm and the non-robust version of the algorithm. Videos of the experiments are shown at https://***/view/rfvi Copyright © 2021, The Authors. All rights reserved.

关键词： Dynamic programming