检索结果-内蒙古大学图书馆

Fribo: A Social Networking Robot for Increasing Social Connectedness through Sharing Daily Home Activities from Living Noise Data

学校读者我要写书评

暂无评论

Fribo: A Social Networking Robot for Increasing Social Conne...

ACM/IEEE International Conference on human-Robot interaction (HRI)

作者： Kwangmin Jeong Jihyun Sung Haesung Lee Aram Kim Hyemi Kim Chanmi Park Youin Jeong JeeHang Lee Jinwoo Kim Human-Computer Interaction Lab Yonsei University Seoul Korea Brain and Machine Intelligence Lab KI for Health Science and Technology KAIST Daejeon Korea

The rapid increase in the number of young adults living alone gives rise to a demand for the resolution of social isolation problems. Social robot technologies play a substantial role for this purpose. However, existing technologies try to solve the problem only through one-to-one interaction with robots, which in turn fails to utilize the real-world social relationships. Privacy concern is an additional issue since most social robots rely on the visual information for the interactions. To this end, we propose ‘Fribo’, auditory information centered social robot that recognizes user’s activity by analyzing occupants’ living noise and shares the activity information with close friends. A four-week field study with the first prototype of Fribo confirms that activity sharing through the use of anonymized living noise promises a virtual cohabiting experience that triggers more frequent real-world social interactions with less feeling of privacy intrusion. Based on this finding and the further qualitative analysis, we suggest a design principle of sound-based social networking robots and its associated new interactions, then present the second prototype of Fribo inspired by the implications from the field study.

关键词： Privacy Visualization Social networking (online) Prototypes human-robot interaction Robots

OmniTrack: A flexible self-tracking approach leveraging semi-automated tracking

学校读者我要写书评

暂无评论

Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies 2017年第3期1卷 1-28页

作者： Kim, Young-Ho Jeon, Jae Ho Lee, Bongshin Choe, Eun Kyoung Seo, Jinwook Human-Computer Interaction Lab Dept. of Computer Science and Engineering Seoul National University Seoul Korea Republic of Kakao Corporation Seongnam Gyeonggi-do Korea Republic of Microsoft Research RedmondWA United States Human-Computer Interaction Lab College of Information Studies University of Maryland College ParkMD United States

We now see an increasing number of self-tracking apps and wearable devices. Despite the vast number of available tools, however, it is still challenging for self-trackers to find apps that suit their unique tracking needs, preferences, and commitments. Furthermore, people are bounded by the tracking tools' initial design because it is difficult to modify, extend, or mash up existing tools. In this paper, we present OmniTrack, a mobile self-tracking system, which enables self-trackers to construct their own trackers and customize tracking items to meet their individual tracking needs. To inform the OmniTrack design, we first conducted semi-structured interviews (N = 12) and analyzed existing mobile tracking apps (N = 62). We then designed and developed OmniTrack as an Android mobile app, leveraging a semi-automated tracking approach that combines manual and automated tracking methods. We evaluated OmniTrack through a usability study (N = 10) and improved its interfaces based on the feedback. Finally, we conducted a 3-week deployment study (N = 21) to assess if people can capitalize on OmniTrack's flexible and customizable design to meet their tracking needs. From the study, we showed how participants used OmniTrack to create, revise, and appropriate trackers-ranging from a simple mood tracker to a sophisticated daily activity tracker. We discuss how OmniTrack positively influences and supports self-trackers' tracking practices over time, and how to further improve OmniTrack by providing more appropriate visualizations and sharable templates, incorporating external contexts, and supporting researchers' unique data collection needs. © 2017 Copyright held by the owner/author(s). Publication rights licensed to Association for Computing Machinery.

关键词： Automation

Controllable dual skew divergence loss for neural machine translation

学校读者我要写书评

暂无评论

arXiv 2019年

作者： Li, Zuchao Zhao, Hai Wu, Yingting Xiao, Fengshun Jiang, Shu Department of Computer Science and Engineering Shanghai Jiao Tong University China Key Laboratory of Shanghai Education Commission for Intelligent Interaction and Cognitive Engineering Shanghai Jiao Tong University China MoE Key Lab of Artificial Intelligence AI Institute Shanghai Jiao Tong University China

In sequence prediction tasks like neural machine translation, training with cross-entropy loss often leads to models that overgeneralize and plunge into local optima. In this paper, we propose an extended loss function called dual skew divergence (DSD) that integrates two symmetric terms on KL divergences with a balanced weight. We empirically discovered that such a balanced weight plays a crucial role in applying the proposed DSD loss into deep models. Thus we eventually develop a controllable DSD loss for general-purpose scenarios. Our experiments indicate that switching to the DSD loss after the convergence of ML training helps models escape local optima and stimulates stable performance improvements. Our evaluations on the WMT 2014 English-German and English-French translation tasks demonstrate that the proposed loss as a general and convenient mean for NMT training indeed brings performance improvement in comparison to strong baselines. Copyright © 2019, The Authors. All rights reserved.

关键词： Neural machine translation

学校读者我要写书评

暂无评论

Subword ELMo

arXiv 2019年

作者： Li, Jiangtong Zhao, Hai Li, Zuchao Bi, Wei Liu, Xiaojiang Department of Computer Science and Engineering Shanghai Jiao Tong University Key Laboratory of Shanghai Education Commission for Intelligent Interaction and Cognitive Engineering Shanghai Jiao Tong University Shanghai China MoE Key Lab of Artificial Intelligence AI Institute Shanghai Jiao Tong University Tencent AI Lab Shenzhen China

Embedding from Language Models (ELMo) has shown to be effective for improving many natural language processing (NLP) tasks, and ELMo takes character information to compose word representation to train language models. However, the character is an insufficient and unnatural linguistic unit for word representation. Thus we introduce Embedding from Subword-aware Language Models (ESuLMo) which learns word representation from subwords using unsupervised segmentation over words. We show that ESuLMo can enhance four benchmark NLP tasks more effectively than ELMo, including syntactic dependency parsing, semantic role labeling, implicit discourse relation recognition and textual entailment, which brings a meaningful improvement over ELMo. Copyright © 2019, The Authors. All rights reserved.

关键词： Embeddings

Exploration of Interpretability Techniques for Deep COVID-19 Classification using Chest X-ray Images

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Chatterjee, Soumick Saad, Fatima Sarasaen, Chompunuch Ghosh, Suhita Krug, Valerie Khatun, Rupali Mishra, Rahul Desai, Nirja Radeva, Petia Rose, Georg Stober, Sebastian Speck, Oliver Nürnberger, Andreas Data and Knowledge Engineering Group Otto von Guericke University Magdeburg Germany Faculty of Computer Science Otto von Guericke University Magdeburg Germany Genomics Research Centre Human Technopole Italy Institute for Medical Engineering Otto von Guericke University Magdeburg Germany Research Campus STIMULATE Otto von Guericke University Magdeburg Germany Biomedical Magnetic Resonance Otto von Guericke University Magdeburg Germany Artificial Intelligence Lab Otto von Guericke University Magdeburg Germany Department of Mathematics and Computer Science University of Barcelona Barcelona Spain Translational Radiobiology Department of Radiation Oncology Universitätsklinikum Erlangen Erlangen Germany Apollo Hospitals Bilaspur India HCG Cancer Centre Vadodara India Computer Vision Centre Cerdanyola Barcelona Spain Centre for Behavioural Brain Sciences Magdeburg Germany German Centre for Neurodegenerative Diseases Magdeburg Germany

The outbreak of COVID-19 has shocked the entire world with its fairly rapid spread and has challenged different sectors. One of the most effective ways to limit its spread is the early and accurate diagnosing infected patients. Medical imaging, such as X-ray and Computed Tomography (CT), combined with the potential of Artificial Intelligence (AI), plays an essential role in supporting medical personnel in the diagnosis process. Thus, in this article five different deep learning models (ResNet18, ResNet34, InceptionV3, InceptionResNetV2 and DenseNet161) and their ensemble, using majority voting have been used to classify COVID-19, pneumoniæ and healthy subjects using chest X-ray images. Multilabel classification was performed to predict multiple pathologies for each patient, if present. Firstly, the interpretability of each of the networks was thoroughly studied using local interpretability methods - occlusion, saliency, input X gradient, guided backpropagation, integrated gradients, and DeepLIFT, and using a global technique - neuron activation profiles. The mean Micro-F1 score of the models for COVID-19 classifications ranges from 0.66 to 0.875, and is 0.89 for the ensemble of the network models. The qualitative results showed that the ResNets were the most interpretable models. This research demonstrates the importance of using interpretability methods to compare different models before making a decision regarding the best performing model. Copyright © 2020, The Authors. All rights reserved.

关键词： COVID-19

Probing contextualized sentence representations with visual awareness

学校读者我要写书评

暂无评论

arXiv 2019年

作者： Zhang, Zhuosheng Wang, Rui Chen, Kehai Utiyama, Masao Sumita, Eiichiro Zhao, Hai Department of Computer Science and Engineering Shanghai Jiao Tong University Key Laboratory of Shanghai Education Commission for Intelligent Interaction and Cognitive Engineering Shanghai Jiao Tong University Shanghai China MoE Key Lab of Artificial Intelligence AI Institute Shanghai Jiao Tong University Shanghai China

We present a universal framework to model contextualized sentence representations with visual awareness that is motivated to overcome the shortcomings of the multimodal parallel data with manual annotations. For each sentence, we first retrieve a diversity of images from a shared cross-modal embedding space, which is pre-trained on a large-scale of text-image pairs. Then, the texts and images are respectively encoded by transformer encoder and convolutional neural network. The two sequences of representations are further fused by a simple and effective attention layer. The architecture can be easily applied to text-only natural language processing tasks without manually annotating multimodal parallel corpora. We apply the proposed method on three tasks, including neural machine translation, natural language inference and sequence labeling and experimental results verify the effectiveness. Copyright © 2019, The Authors. All rights reserved.

关键词： Natural language processing systems

A Smart Sliding Chinese Pinyin Input Method Editor on Touchscreen

学校读者我要写书评

暂无评论

arXiv 2019年

作者： Zhang, Zhuosheng Meng, Zhen Zhao, Hai Department of Computer Science and Engineering Shanghai Jiao Tong University Key Laboratory of Shanghai Education Commission for Intelligent Interaction and Cognitive Engineering Shanghai Jiao Tong University Shanghai China MoE Key Lab of Artificial Intelligence AI Institute Shanghai Jiao Tong University Shanghai China

This paper presents a smart sliding Chinese pinyin Input Method Editor (IME) for touchscreen devices which allows user finger sliding from one key to another on the touchscreen instead of tapping keys one by one, while the target Chinese character sequence will be predicted during the sliding process to help user input Chinese characters efficiently. Moreover, the layout of the virtual keyboard of our IME adapts to user sliding for more efficient inputting. The layout adaption process is utilized with Recurrent Neural Networks (RNN) and deep reinforcement learning. The pinyin-to-character converter is implemented with a sequence-to-sequence (Seq2Seq) model to predict the target Chinese sequence. A sliding simulator is built to automatically produce sliding samples for model training and virtual keyboard test. The key advantage of our proposed IME is that nearly all its built-in tactics can be optimized automatically with deep learning algorithms only following user behavior. Empirical studies verify the effectiveness of the proposed model and show a better user input efficiency. Copyright © 2019, The Authors. All rights reserved.

关键词： Behavioral research

AGlobal benchmark of algorithms for segmentinglate gadolinium-enhanced cardiac magnetic resonance imaging

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Xiong, Zhaohan Xia, Qing Hu, Zhiqiang Huang, Ning Bian, Cheng Zheng, Yefeng Vesal, Sulaiman Ravikumar, Nishant Maier, Andreas Yang, Xin Heng, Pheng-Ann Ni, Dong Li, Caizi Tong, Qianqian Si, Weixin Puybareau, Elodie Khoudli, Younes Géraud, Thierry Chen, Chen Bai, Wenjia Rueckert, Daniel Xu, Lingchao Zhuang, Xiahai Luo, Xinzhe Jia, Shuman Sermesant, Maxime Liu, Yashu Wang, Kuanquan Borra, Davide Masci, Alessandro Corsi, Cristiana De Vente, Coen Veta, Mitko Karim, Rashed Preetha, Chandrakanth Jayachandran Engelhardt, Sandy Qiao, Menyun Wang, Yuanyuan Tao, Qian Nuñez-Garcia, Marta Camara, Oscar Savioli, Nicolo Lamata, Pablo Zhao, Jichao Auckland Bioengineering Institute University of Auckland Auckland New Zealand State Key Lab of Virtual Reality Technology and Systems Beihang University Beijing China School of Electronics Engineering and Computer Science Peking University Beijing China SenseTime Inc Shenzhen China Tencent Jarvis Laboratory Shenzhen China Pattern Recognition Lab Friedrich-Alexander-Universität Erlangen-Nürnberg Erlangen Germany Department of Computer Science and Engineering Chinese University of Hong Kong Hong Kong National-Regional Key Technology Engineering Laboratory for Medical Ultrasound School of Biomedical Engineering Shenzhen University Shenzhen China School of Computer Science Wuhan University Wuhan China Shenzhen Key Laboratory of Virtual Reality and Human Interaction Technology Shenzhen Institutes of Advanced Technology Chinese Academy of Sciences Shenzhen China Epita Research and Development Laboratory Paris France Department of Computing Imperial College London London United Kingdom School of Naval Architecture Shanghai Jiao Tong University Ocean & Civil Engineering Shanghai China School of Data Science Fudan University Shanghai China Epione Research Group Universite Cote d'Azur Inria Sophia Antipolis France Harbin Institute of Technology School of Computer Science and Technology Harbin China Department of Electric University of Bologna Electronic and Information Engineering Cesena Italy Department of Biomedical Engineering Eindhoven University of Technology Eindhoven Netherlands School of Biomedical Engineering and Imaging Sciences Kings College London London United Kingdom Faculty of Electrical Engineering and Information Technology University of Magdeburg Magdeburg Germany Heidelberg University Hospital Germany Biomedical Engineering Center Fudan University Shanghai China Department of Radiology Leiden University Medical Center Leiden Netherlands Department of Information and Communication Technologies Universitat Pompeu Fabra Barcelona Spain D

Segmentation of cardiac images, particularly late gadolinium-enhanced magnetic resonance imaging (LGE-MRI)widely used for visualizing diseased cardiacstructures, is a crucial first step for clinical diagnosis and treatment. However, direct segmentation ofLGE-MRIsischallenging due to *** most clinical studies have relied on manual and labor-intensive approaches, automatic methods are of high interest, particularly optimized machine learning *** address this, we organized the "2018 Left Atrium Segmentation Challenge" using 154 3D LGE-MRIs, currently the world's largest cardiac LGE-MRI dataset, and associated labels of the left atriumsegmented by three medical experts,ultimately attractingthe participation of27 international teams. In this paper, extensive analysis of the submitted algorithms using technical and biological metrics was performed by undergoing subgroup analysis and conducting hyper-parameter analysis,offeringan overall picture ofthe major design choices of convolutional neural networks(CNNs) and practical considerations for achieving state-of-the-art left atrium segmentation. Results show the top methodachieved a dice score of 93.2%and a mean surface to a surface distance of 0.7mm, significantly outperforming prior state-of-the-art. Particularly, our analysisdemonstrated that double, sequentially usedCNNs, in which a first CNN is used for automatic region-of-interestlocalization and a subsequent CNN is used for refinedregional segmentation, achieved far superiorresults than traditional methods and pipelinescontaining *** findings from this study can potentially be extended to other imaging datasets and modalities, having an impact on the wider medical imaging *** large-scale benchmarking study makes a significant step towards much-improved segmentation methods for cardiac LGE-MRIs, and will serveas animportant benchmark for evaluating and comparing the future works in the field. Copyright © 2020, Th

关键词： Magnetic resonance imaging

EEG Data Augmentation for Emotion Recognition Using a Conditional Wasserstein GAN 40

学校读者我要写书评

暂无评论

EEG Data Augmentation for Emotion Recognition Using a Condit...

40th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBC 2018

作者： Luo, Yun Center for Brain-Like Computing and Machine Intelligence Department of Computer Science and Engineering Key Lab. of Shanghai Education Commission for Intelligent Interaction and Cognitive Engineering Brain Science and Technology Research Center Shanghai Jiao Tong University 800 Dong Chuan Road Shanghai200240 China

ISBN: (纸本)9781538636466

Due to the lack of electroencephalography (EEG) data, it is hard to build an emotion recognition model with high accuracy from EEG signals using machine learning approach. Inspired by generative adversarial networks (GANs), we introduce a Conditional Wasserstein GAN (CWGAN) framework for EEG data augmentation to enhance EEG-based emotion recognition. A Wasserstein GAN with gradient penalty is adopted to generate realistic-like EEG data in differential entropy (DE) form. Three indicators are used to judge the qualities of the generated high-dimensional EEG data, and only high quality data are appended to supplement the data manifold, which leads to better classification of different emotions. We evaluate the proposed CWGAN framework on two public EEG datasets for emotion recognition, namely SEED and DEAP. The experimental results demonstrate that using the EEG data generated by CWGAN significantly improves the accuracies of emotion recognition models. © 2018 IEEE.

关键词： Electroencephalography Gallium nitride Training Generators Emotion recognition Generative adversarial networks Brain modeling