检索结果-内蒙古大学图书馆

Sub-architecture ensemble pruning in neural architecture search

学校读者我要写书评

暂无评论

arXiv 2019年

作者： Bian, Yijun Song, Qingquan Du, Mengnan Yao, Jun Chen, Huanhuan Hu, Xia The School of Computer Science and Technology University of Science and Technology of China Hefei230027 China The Department of Computer Science and Engineering Texas A&M University College StationTX77840 United States The Data Science and Analytics Department WeBank Shenzhen518000 China

Neural architecture search (NAS) is gaining more and more attention in recent years due to its flexibility and remarkable capability to reduce the burden of neural network design. To achieve better performance, however, the searching process usually costs massive computations that might not be affordable for researchers and practitioners. While recent attempts have employed ensemble learning methods to mitigate the enormous computational cost, however, they neglect a key property of ensemble methods, namely diversity, which leads to collecting more similar sub-architectures with potential redundancy in the final design. To tackle this problem, we propose a pruning method for NAS ensembles called "Sub-Architecture Ensemble Pruning in Neural Architecture Search (SAEP)." It targets to leverage diversity and to achieve sub-ensemble architectures at a smaller size with comparable performance to ensemble architectures that are not pruned. Three possible solutions are proposed to decide which sub-architectures to prune during the searching process. Experimental results exhibit the effectiveness of the proposed method by largely reducing the number of sub-architectures without degrading the performance. Copyright © 2019, The Authors. All rights reserved.

关键词： Network architecture

Entanglement induced barren plateaus

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Marrero, Carlos Ortiz Kieferová, Mária Wiebe, Nathan Data Sciences and Analytics Group Pacific Northwest National Laboratory RichlandWA99354 United States Centre for Quantum Computation and Communication Technology Centre for Quantum Software and Information University of Technology Sydney NSW2007 Australia Department of Computer Science University of Toronto ONM5S 1A1 Canada

We argue that an excess in entanglement between the visible and hidden units in a Quantum Neural Network can hinder learning. In particular, we show that quantum neural networks that satisfy a volume-law in the entanglement entropy will give rise to models not suitable for learning with high probability. Using arguments from quantum thermodynamics, we then show that this volume law is typical and that there exists a barren plateau in the optimization landscape due to entanglement. More precisely, we show that for any bounded objective function on the visible layers, the Lipshitz constants of the expectation value of that objective function will scale inversely with the dimension of the hidden-subsystem with high probability. We show how this can cause both gradient descent and gradient-free methods to fail. We note that similar problems can happen with quantum Boltzmann machines, although stronger assumptions on the coupling between the hidden/visible subspaces are necessary. We highlight how pretraining such generative models may provide a way to navigate these barren plateaus. Copyright © 2020, The Authors. All rights reserved.

关键词： Gradient methods

Enhancing decision making capacity in tourism domain using social media analytics 18

学校读者我要写书评

暂无评论

Enhancing decision making capacity in tourism domain using s...

18th International Conference on Advances in ICT for Emerging Regions, ICTer 2018

作者： Abeysinghe, Supun Manchanayake, Isura Samarajeewa, Chamod Rathnayaka, Prabod Walpola, Malaka J. Nawaratne, Rashmika Bandaragoda, Tharindu Alahakoon, Damminda Department of Computer Science and Engineering University of Moratuwa Sri Lanka Research Centre for Data Analytics and Cognition La Trobe University Victoria Australia

ISBN: (纸本)9781538673515

Social media has gained an immense popularity over the last decade. People tend to express opinions about their daily encounters on social media freely. These daily encounters include the places they travelled, hotels or restaurants they have tried and aspects related to tourism in general. Since people usually express their true experiences on social media, the expressed opinions contain valuable information that can be used to generate business value and aid in decision-making processes. Due to the large volume of data, it is not a feasible task to manually go through each and every item and extract the information. Hence, we propose a social media analytics platform which has the capability to identify discussion pathways and aspects with their corresponding sentiment and deeper emotions using machine learning techniques and a visualization tool which shows the extracted insights in a comprehensible and concise manner. Identified topic pathways and aspects will give a decision maker some insight into what are the most discussed topics about the entity whereas associated sentiments and emotions will help to identify the feedback. © 2018 IEEE.

关键词： Machine learning

Informal caregivers' attitudes and compliance towards a connected health platform for home care support: Insights from a long-term exposure

学校读者我要写书评

暂无评论

Gerontechnology 2019年第4期18卷 231-242页

作者： Guisado-Fernández, Estefanía Giunti, Guido Mackey, Laura Silva, Paula Alexandra Blake, Catherine Caulfield, Brian Insight Centre for Data Analytics University College Dublin 3rd floor O'Brien Science Building East Dublin Ireland M3S Research Unit Faculty of Information Technology and Electrical Engineering University of Oulu Oulu Finland School of Public Health Physiotherapy and Sports Science University College Dublin Dublin Ireland Department of Computer Science Center for Informatics and Systems of the University of Coimbra (CISUC) University of Coimbra Coimbra Portugal

Background When designing Connected Health (CH) solutions for home care, it is vital to focus on usability and user experience to ensure that technologies are easy to use and meet users' expectations and needs. Generally, the usability and user experience tests are conducted during short-term exposure, which does not allow a true understanding of how the technology can help with the home caring tasks. Research aim We aim to investigate informal caregivers' feedback on the utility and usability of a CH monitoring platform for People with Dementia (PwD) during a period of extended use in the natural living context, and to understand how this was related to compliance patterns. Methods Informal caregiver's feedback about the CH platform, usability, and the impact of short-term versus long-term exposure were investigated through semi-structured individual interviews at the beginning and end of a 6-month deployment in the home care setting. Informal caregivers' compliance with the CH platform was analysed from their daily platform utilization during the deployment time. Results 11 informal caregivers agreed to participate. There was a change in the participants' opinions about the CH platform between the short-term and the long-term exposure feedback. Their initial impressions about what the CH platform could offer them to improve their delivery of home care for the PwD did not correspond with what they found that the CH platform could provide them following the long-term exposure. If at the beginning they saw the CH platform as a helpful tool to facilitate home care delivery and to improve their self-efficacy, after the deployment they expressed that because of the way the platform was designed it was mainly conceived for dementia research benefit and not to fulfill their caring needs. Compliance with the CH platform was quite low and similar between all participants. Conclusions Most contemporary CH studies are not conducted in real-life settings and without enough dur

关键词： Connected health Dementia Informal caregiver Usability User experience

Storywrangler: A massive exploratorium for sociolinguistic, cultural, socioeconomic, and political timelines using Twitter

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Alshaabi, Thayer Adams, Jane L. Arnold, Michael V. Minot, Joshua R. Dewhurst, David R. Reagan, Andrew J. Danforth, Christopher M. Dodds, Peter Sheridan Vermont Complex Systems Center Computational Story Lab University of Vermont BurlingtonVT05405 United States Department of Computer Science University of Vermont BurlingtonVT05405 United States Charles River Analytics CambridgeMA02138 United States MassMutual Data Science AmherstMA01002 United States Department of Mathematics & Statistics University of Vermont BurlingtonVT05405 United States

In real-time, Twitter strongly imprints world events, popular culture, and the day-to-day, recording an ever growing compendium of language change. Vitally, and absent from many standard corpora such as books and news archives, Twitter also encodes popularity and spreading through retweets. Here, we describe Storywrangler, an ongoing curation of over 100 billion tweets containing 1 trillion 1-grams from 2008 to 2021. For each day, we break tweets into 1-, 2-, and 3-grams across 100+ languages, generating frequencies for words, hashtags, handles, numerals, symbols, and emojis. We make the dataset available through an interactive time series viewer, and as downloadable time series and daily distributions. Although Storywrangler leverages Twitter data, our method of tracking dynamic changes in n-grams can be extended to any temporally evolving corpus. Illustrating the instrument’s potential, we present example use cases including social amplification, the sociotechnical dynamics of famous individuals, box office success, and social unrest. Copyright © 2020, The Authors. All rights reserved.

关键词： Social networking (online)

Urban sensing based on mobile phone data: Approaches, applications and challenges

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Ghahramani, Mohammadhossein Zhou, MengChu Wang, Gang Insight Centre for Data Analytics University College Dublin Ireland Institute of Systems Engineering Macau University of Science and Technology Macau999078 China China Telecom at Macau Macau999078 China Institute of Systems Engineering Macau University of Science and Technology Macau999078 China Department of Electrical and Computer Engineering New Jersey Institute of Technology NewarkNJ07102 United States

data volume grows explosively with the proliferation of powerful smartphones and innovative mobile applications. The ability to accurately and extensively monitor and analyze these data is necessary. Much concern in mobile data analysis is related to human beings and their behaviours. Due to the potential value that lies behind these massive data, there have been different proposed approaches for understanding corresponding patterns. To that end, monitoring people’s interactions, whether counting them at fixed locations or tracking them by generating origindestination matrices is crucial. The former can be used to determine the utilization of assets like roads and city attractions. The latter is valuable when planning transport infrastructure. Such insights allow a government to predict the adoption of new roads, new public transport routes, modification of existing infrastructure, and detection of congestion zones, resulting in more efficient designs and improvement. Smartphone data exploration can help research in various fields, e.g., urban planning, transportation, health care, and business marketing. It can also help organizations in decision making, policy implementation, monitoring and evaluation at all levels. This work aims to review the methods and techniques that have been implemented to discover knowledge from mobile phone data. We classify these existing methods and present a taxonomy of the related work by discussing their pros and cons. Copyright © 2020, The Authors. All rights reserved.

关键词： Smartphones

Semi-unsupervised lifelong learning for sentiment classification: Less manual data annotation and more self-studying

学校读者我要写书评

暂无评论

arXiv 2019年

作者： Hong, Xianbin Wong, Prudence Pal, Gautam Liu, Dawei Huang, Xin Guan, Sheng-Uei Man, Ka Lok Research Institute of Big Data Analytics Xi'an Jiaotong-Liverpool University Suzhou Jiangsu China Department of Computer Science University of Liverpool Liverpool Merseyside United Kingdom

Lifelong machine learning is a novel machine learning paradigm which can continually accumulate knowledge during learning. The knowledge extracting and reusing abilities enable the lifelong machine learning to solve the related problems. The traditional approaches like Naïve Bayes and some neural network based approaches only aim to achieve the best performance upon a single task. Unlike them, the lifelong machine learning in this paper focus on how to accumulate knowledge during learning and leverage them for the further tasks. Meanwhile, the demand for labeled data for training also be significantly decreased with the knowledge reusing. This paper suggests that the aim of the lifelong learning is to use less labeled data and computational cost to achieve the performance as well as or even better than the supervised learning. Copyright © 2019, The Authors. All rights reserved.

关键词： Machine learning

UP-CNN: Un-pooling augmented convolutional neural network

学校读者我要写书评

暂无评论

Pattern Recognition Letters 2019年 119卷 34-40页

作者： Chunyan Xu Jian Yang Hanjiang Lai Junbin Gao Linlin Shen Shuicheng Yan School of Computer Science and Engineering Nanjing University of Science and Technology 210094 China School of Data and Computer Science Sun Yat-Sen University 510275 China Discipline of Business Analytics University of Sydney Business School Universtiy of Sydney NSW 2006 Australia Computer Vision Institute School of Computer Science and Software Engineering Shenzhen University 518060 China Department of Electrical and Computer Engineering National University of Singapore 117583 Singapore

Convolutional neural network (CNN) has shown remarkable performance in various visual recognition tasks. Most of existing CNN is a purely bottom-up and feed-forward architecture, we argue that it fails to consider the interaction between low-level fine details and high-level semantic information . In this paper, a novel “Un-Pooling augmented Convolutional Neural Network” (UP-CNN) is proposed to boost the discriminative capability of the CNN with the following three distinctive properties: (1) UP-CNN is a deeper network, which is comprised of a bottom-up, a top-down and then a bottom-up sub-networks, associating to different level information that jointly improves its discriminative capability. (2) With the mixture of pooling and un-pooling layers, UP-CNN easily allows the interaction across convolutional layers with the same size from different sub-networks. This architecture effectively depresses the attenuation of important information including both activations in the forward process and gradient information in the back-propagation process. (3) UP-CNN employs the ratio un-pooling operation to reconstruct activations of the original size in the top-down sub-network, where the spatial information that is lost during pooling can be preserved within a receptive field. The experiments on four benchmark datasets (including the CIFAR-10, CIFAR-100, MNIST and SVHN datasets) well demonstrate that the proposed UP-CNN architecture considerably outperforms other state-of-the-art methods.

关键词：

Benchmarking bias: Expanding clinical AI model card to incorporate bias reporting of social and non-social factors

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Heming, Carolina A.M. Abdalla, Mohamed Mohanna, Shahram Ahluwalia, Monish Zhang, Linglin Trivedi, Hari Woo, MinJae Fine, Benjamin Gichoya, Judy Wawira Celi, Leo Anthony Seyyed-Kalantari, Laleh University of Iowa Iowa CityIA United States Institute for Better Health Trillium Health Partners 100 Queensway West 6th floor MississaugaONL5B 1B8 Canada Kingston Health Sciences Centre Queen's University 76 Stuart St. KingstonONK7L 2V7 Canada School of Data Science and Analytics Kennesaw State University 3391 Town Point Dr NW KennesawGA30144 United States Department of Radiology and Imaging Sciences Emory University 1364 E Clifton Rd NE AtlantaGA30322 United States Vector Institute of Artificial Intelligence 661 University Ave Suite 710 TorontoONM5G 1M1 Canada Massachusetts Institute of Technology 77 Massachusetts Avenue CambridgeMA02139 United States Harvard Medical School 25 Shattuck St BostonMA02115 United States Department of Electrical Engineering and Computer Science York University 4700 Keele St TorontoONM3J 1P3 Canada Department of Medicine University of Alberta Walter C Mackenzie Health Sciences Centre St. NW Edmonton Alberta 8440 112 Canada

The model facts labels are often used to communicate artificial intelligence (AI) model performance to the end users for proper deployment and often report the overall performance averaged across all populations. However, the overall performance is not generalizable to different subgroups of patients. In this paper, we highlight the importance of currently unforeseen aspects, essential for AI models in healthcare, that are not included in the model fact cards. We suggest including through bias and fairness analysis outcomes in model fact labels in the health care AI model including: (a) socio-demographic fairness factors (e.g., sexes, race, socioeconomic status, etc.) and (b) Technical, Environmental and Biological Bias Factors (e.g. disease-dependent, anatomical, or instrumental factors, etc.). The socio-demographic fairness factors have been explored in fairness analysis of AI models in healthcare but have not often been listed in model fact labels. Besides, the technical, environmental, and biological bias factors have not often been explored in the bias analysis of AI models in healthcare and have not been reported in model fact cards. Here, we demonstrate disparities with respect to both sets of factors. We suggest that clinical AI model reporting cards need to be expanded to communicate such a wider range of disparities to end users for precise patient care. Analysis and communication of such a wider range of biases identify potential drivers of bias, which pave the way to debiasing. As well, it highlights the model caveats and guides the end user and policymaker on how, when, and under what circumstances the model works or fails. Copyright © 2023, The Authors. All rights reserved.

关键词：