基于注意力机制和新型真值推测技术的智能环境温度感知系统

监控室内温度可以更好地节约能源并提高舒适度。作为普及性极高且搭载多种传感器的移动设备，智能手机成为了理想的测温工具选择。其温度预测结果可以作为额外数据源，帮助校准电子温度计的读数。然而，由于使用状态的差异，智能手机的预测结果有时会产生较大误差。同时，在众包问题领域，目前提出的真值推测模型很少考虑标注结果的可信度。这可能是因为同时收集答案和参与者对答案的确信程度会带来高昂成本并引入主观性。为此，本研究首先构建了一个包含多元数据以及多个场景的手机温度预测数据集。基于该数据集，本文提出了一个包含两个部分模型的环境温度感知系统。第一部分的温度预测模型利用智能手机上易获取的状态信息数据作为输入，输出周围环境温度的预测结果及其对应的置信度。第二部分的真值推测模型则读取多个预测结果和相应生成的置信度，输出推测的真实结果。单个手机的温度预测模型在本文构建的数据集上表现出了令人满意的测温准确性，平均绝对误差达到了0.253℃。同时，本研究通过迁移学习验证了该测温模型适用于各种型号的智能手机。基于置信度的真值推测模型在测试集上取得了0.128℃的平均绝对误差，明显优于目前许多真值推测模型的结果，并且相较于单台手机，测温精度大幅提升。我们相信本研究将对能源节约和众包方法提供新的启示。

Monitoring the indoor temperature can help saving of energy and improve the comfort level. Smartphone, as a ubiquitous device, can be an additional data source to provide the ambient temperature estimation. However, the estimation results sometimes can be unreliable due to the different phone using states. At the same time, the state-of-the-art crowdsourcing truth inference algorithms rarely consider the confidence of the given answers, which, may because of high subjectivity and high collection difficulty of confidence. In this work, firstly, we construct a multi-scenario mobile phone dataset for the work of mobile phone temperature measurement. We also proposed one phone-based ambient temperature measurement system which contains two models. The first prediction model takes easily accessible phone state features as inputs and outputs ambient temperature prediction with the confidence level. The second truth inference model takes multiple prediction results with confidences as inputs and outputs a referred final answer. We evaluate our temperature estimation model in testing dataset and it reaches 0.253 degrees Celsius with MAE (Mean Absolute Error). We also proved by transfer learning our model can be used in a new type of phone. We evaluate the truth inference model in our testing dataset and reaches 0.128 degrees Celsius, which outperforms the state-of-the-art truth inference algorithms. We believe this work can contribute to energy conservation and provide new ideas for crowdsourcing.

[1]ALLOUHI A, EL FOUIH Y, KOUSKSOU T, et al. Energy consumption and efficiency in buildings: current status and future trends[J]. Journal of Cleaner production, 2015, 109: 118- 130.
[2]DARBY S, et al. The effectiveness of feedback on energy consumption[J]. A Review for DEFRA of the Literature on Metering, Billing and direct Displays, 2006, 486(2006): 26.
[3]PÉREZ-LOMBARD L, ORTIZ J, POUT C. A review on buildings energy consumption information[J]. Energy and buildings, 2008, 40(3): 394-398.
[4]TRIVEDI A, BOVORNKEERATIROJ P, BREDA J, et al. Phone-based ambient temperature sensing using opportunistic crowdsensing and machine learning[J]. Sustainable Computing: Informatics and Systems, 2021, 29: 100479.
[5]CHEN S, WANG J, ZHANG L, et al. When internet of things meets e-health: an indoor temperature monitoring and control approach[J]. IEEE Internet of Things Magazine, 2021, 4(3): 12-16.
[6]CHEN S, ZHANG L, TANG Y, et al. Indoor temperature monitoring using wireless sensor networks: a SMAC application in smart cities[J]. Sustainable Cities and Society, 2020, 61: 102333.
[7]THAM S, THOMPSON R, LANDEG O, et al. Indoor temperature and health: a global systematic review[J]. Public Health, 2020, 179: 9-17.
[8]TRIVEDI A, BOVORNKEERATIROJ P, BREDA J, et al. Phone-based ambient temperature sensing using opportunistic crowdsensing and machine learning[J]. Sustainable Computing: Informatics and Systems, 2021, 29: 100479.
[9]SONG K, LIU X, GAO T. Potential application of using smartphone sensor for estimating air temperature: experimental study[J]. IEEE Internet of Things Journal, 2021, 9(16): 14300- 14306.
[10]DROSTE A, PAPE J J, OVEREEM A, et al. Crowdsourcing urban air temperatures through smartphone battery temperatures in São Paulo, Brazil[J]. Journal of Atmospheric and Oceanic Technology, 2017, 34(9): 1853-1866.
[11]OVEREEM A, R. ROBINSON J, LEIJNSE H, et al. Crowdsourcing urban air temperatures from smartphone battery temperatures[J]. Geophysical Research Letters, 2013, 40(15): 4081-4085.
[12]BREDA J, TRIVEDI A, WIJESUNDARA C, et al. Hot or not: Leveraging mobile devices for ubiquitous temperature sensing[C]//Proceedings of the 6th ACM International Conference on Systems for Energy-Efficient Buildings, Cities, and Transportation. 2019: 41-50.
[13]DICKINSON J L, ZUCKERBERG B, BONTER D N. Citizen science as an ecological research tool: challenges and benefits[J]. Annual review of ecology, evolution, and systematics, 2010, 41: 149-172.
[14]ZHANG H, CHEN J, YAN J, et al. Urban power load profiles under ageing transition integrated with future EVs charging[J]. Advances in Applied Energy, 2021, 1: 100007.
[15]ZHANG H, YAN J, YU Q, et al. 1.6 Million transactions replicate distributed PV market slowdown by COVID-19 lockdown[J]. Applied Energy, 2021, 283: 116341.
[16]SUI Y, ZHANG H, SHANG W, et al. Mining urban sustainable performance: Spatio-temporal emission potential changes of urban transit buses in post-COVID-19 future[J]. Applied Energy, 2020, 280: 115966.
[17]ZHANG H, SONG X, XIA T, et al. Battery electric vehicles in Japan: Human mobile behavior based adoption potential analysis and policy target response[J]. Applied Energy, 2018, 220: 527-535.
[18]CHEN J, LI W, ZHANG H, et al. Mining urban sustainable performance: GPS data-based spatio-temporal analysis on on-road braking emission[J]. Journal of Cleaner Production, 2020, 270: 122489.
[19]JAIN N, SRIVASTAVA V. Data mining techniques: a survey paper[J]. IJRET: International Journal of Research in Engineering and Technology, 2013, 2(11): 2319-1163.
[20]XINTONG G, HONGZHI W, SONG Y, et al. Brief survey of crowdsourcing for data mining [J]. Expert Systems with Applications, 2014, 41(17): 7987-7994.
[21]GAO H, BARBIER G, GOOLSBY R. Harnessing the crowdsourcing power of social media for disaster relief[J]. IEEE intelligent systems, 2011, 26(3): 10-14.
[22]MARKOWSKY G. Crowdsourcing, big data and homeland security[C]//2013 IEEE international conference on technologies for homeland security (HST). IEEE, 2013: 772-778.
[23]ACKERMAN S. Data for the Boston marathon investigation will be crowdsourced[J]. Wired Magazine, 2013, 13.
[24]LIBERT B, SPECTOR J, TAPSCOTT D. We are smarter than me: How to unleash the power of crowds in your business[M]. Pearson Prentice Hall, 2007.
[25]BRABHAM D C, SANCHEZ T W, BARTHOLOMEW K. Crowdsourcing public participation in transit planning: preliminary results from the next stop design case[C]//TRB 89th Annual Meeting Compendium. Citeseer, 2010.
[26]KOLLER D, FRIEDMAN N. Probabilistic graphical models: principles and techniques[M]. MIT press, 2009.
[27]RASMUSSEN C, WILLIAMS C. Gaussian processes for machine learning.,(MIT press: Cambridge, MA)[Z]. 2006.
[28]AYDIN B, YILMAZ Y S Y Y S, LI Y, et al. Crowdsourcing for multiple-choice question answering[C]//Proceedings of the AAAI Conference on Artificial Intelligence: volume 28. 2014: 2946-2953.
[29]WHITEHILL J, WU T F, BERGSMA J, et al. Whose vote should count more: Optimal integration of labels from labelers of unknown expertise[J]. Advances in neural information processing systems, 2009, 22.
[30]MA F, LI Y, LI Q, et al. Faitcrowd: Fine grained truth discovery for crowdsourced data aggregation[C]//Proceedings of the 21th ACM SIGKDD international conference on knowledge discovery and data mining. 2015: 745-754.
[31]FAN J, LI G, OOI B C, et al. icrowd: An adaptive crowdsourcing framework[C]//Proceedings of the 2015 ACM SIGMOD international conference on management of data. 2015: 1015-1030.
[32]DEMARTINI G, DIFALLAH D E, CUDRÉ-MAUROUX P. Zencrowd: leveraging probabilistic reasoning and crowdsourcing techniques for large-scale entity linking[C]//Proceedings of the 21st international conference on World Wide Web. 2012: 469-478.
[33]KARGER D, OH S, SHAH D. Iterative learning for reliable crowdsourcing systems[J]. Advances in neural information processing systems, 2011, 24.
[34]LIU Q, PENG J, IHLER A T. Variational inference for crowdsourcing[J]. Advances in neural information processing systems, 2012, 25.
[35]AYDIN B, YILMAZ Y S Y Y S, LI Y, et al. Crowdsourcing for multiple-choice question answering[C]//Proceedings of the AAAI Conference on Artificial Intelligence: volume 28. 2014: 2946-2953.
[36]DAWID A P, SKENE A M. Maximum likelihood estimation of observer error-rates using the EM algorithm[J]. Journal of the Royal Statistical Society: Series C (Applied Statistics), 1979, 28(1): 20-28.
[37]RAYKAR V C, YU S, ZHAO L H, et al. Learning from crowds.[J]. Journal of machine learning research, 2010, 11(4).
[38]KIM H C, GHAHRAMANI Z. Bayesian classifier combination[C]//Artificial Intelligence and Statistics. PMLR, 2012: 619-627.
[39]VENANZI M, GUIVER J, KAZAI G, et al. Community-based bayesian aggregation models for crowdsourcing[C]//Proceedings of the 23rd international conference on World wide web. 2014: 155-164.
[40]WELINDER P, BRANSON S, PERONA P, et al. The multidimensional wisdom of crowds[J]. Advances in neural information processing systems, 2010, 23.
[41]LI Q, LI Y, GAO J, et al. A confidence-aware approach for truth discovery on long-tail data[J]. Proceedings of the VLDB Endowment, 2014, 8(4): 425-436.
[42]JOGLEKAR M, GARCIA-MOLINA H, PARAMESWARAN A. Evaluating the crowd with confidence[C]//Proceedings of the 19th ACM SIGKDD international conference on Knowledge discovery and data mining. 2013: 686-694.
[43]ZHOU D, BASU S, MAO Y, et al. Learning from the wisdom of crowds by minimax entropy [J]. Advances in neural information processing systems, 2012, 25.
[44]ZHENG Y, LI G, CHENG R. Docs: a domain-aware crowdsourcing system using knowledge bases[J]. Proceedings of the VLDB Endowment, 2016, 10(4): 361-372.
[45]ZHAO Z, WEI F, ZHOU M, et al. Crowd-Selection Query Processing in Crowdsourcing Databases: A Task-Driven Approach.[C]//EDBT. 2015: 397-408.
[46]BERNSTEIN M S, TEEVAN J, DUMAIS S, et al. Direct answers for search queries in the long tail[C]//Proceedings of the SIGCHI conference on human factors in computing systems. 2012: 237-246.
[47]LE J, EDMONDS A, HESTER V, et al. Ensuring quality in crowdsourced search relevance evaluation: The effects of training question distribution[C]//SIGIR 2010 workshop on crowdsourcing for search evaluation: volume 2126. 2010: 22-32.
[48]LI Q, LI Y, GAO J, et al. A confidence-aware approach for truth discovery on long-tail data[J]. Proceedings of the VLDB Endowment, 2014, 8(4): 425-436.
[49]DEMARTINI G, DIFALLAH D E, CUDRÉ-MAUROUX P. Zencrowd: leveraging probabilistic reasoning and crowdsourcing techniques for large-scale entity linking[C]//Proceedings of the 21st international conference on World Wide Web. 2012: 469-478.
[50]CUDRÉ-MAUROUX P, ABERER K, FEHER A. Probabilistic message passing in peer data management systems[C]//22nd International Conference on Data Engineering (ICDE’06). IEEE, 2006: 41-41.
[51]DEMPSTER A P, LAIRD N M, RUBIN D B. Maximum likelihood from incomplete data via the EM algorithm[J]. Journal of the royal statistical society: series B (methodological), 1977, 39(1): 1-22.
[52]WHITEHILL J, WU T F, BERGSMA J, et al. Whose vote should count more: Optimal integration of labels from labelers of unknown expertise[J]. Advances in neural information processing systems, 2009, 22.
[53]KARGER D, OH S, SHAH D. Iterative learning for reliable crowdsourcing systems[J]. Advances in neural information processing systems, 2011, 24.
[54]LIU Q, PENG J, IHLER A T. Variational inference for crowdsourcing[J]. Advances in neural information processing systems, 2012, 25.
[55]DAWID A P, SKENE A M. Maximum likelihood estimation of observer error-rates using the EM algorithm[J]. Journal of the Royal Statistical Society: Series C (Applied Statistics), 1979, 28(1): 20-28.
[56]LI D, REN Z, KANOULAS E. CrowdGP: A Gaussian process model for inferring relevance from crowd annotations[C]//Proceedings of the Web Conference 2021. 2021: 1821-1832.
[57]MORALES-ÁLVAREZ P, RUIZ P, SANTOS-RODRÍGUEZ R, et al. Scalable and efficient learning from crowds with Gaussian processes[J]. Information Fusion, 2019, 52: 110-127.
[58]RODRIGUES F, PEREIRA F, RIBEIRO B. Gaussian process classification and active learning with multiple annotators[C]//International conference on machine learning. PMLR, 2014: 433- 441.
[59]RUIZ P, MORALES-ÁLVAREZ P, MOLINA R, et al. Learning from crowds with variational Gaussian processes[J]. Pattern Recognition, 2019, 88: 298-311.
[60]GROOT P, BIRLUTIU A, HESKES T. Learning from multiple annotators with Gaussian processes[C]//Artificial Neural Networks and Machine Learning–ICANN 2011: 21st International Conference on Artificial Neural Networks, Espoo, Finland, June 14-17, 2011, Proceedings, Part II 21. Springer, 2011: 159-164.
[61]VASWANI A, SHAZEER N, PARMAR N, et al. Attention is all you need[J]. Advances in neural information processing systems, 2017, 30.
[62]ZAREMBA W, SUTSKEVER I, VINYALS O. Recurrent neural network regularization[A]. 2014.
[63]CHOROWSKI J K, BAHDANAU D, SERDYUK D, et al. Attention-based models for speech recognition[J]. Advances in neural information processing systems, 2015, 28.
[64]SHAW P, USZKOREIT J, VASWANI A. Self-attention with relative position representations [A]. 2018.
[65]CHEN Y, DAI X, LIU M, et al. Dynamic convolution: Attention over convolution kernels[C]// Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2020: 11030-11039.
[66]DAI W, YANG Q, XUE G R, et al. Boosting for Transfer Learning[C/OL]//ICML ’07: Proceedings of the 24th International Conference on Machine Learning. New York, NY, USA: Association for Computing Machinery, 2007: 193–200. https://doi.org/10.1145/1273496.1273521.
[67]PAN S J, TSANG I W, KWOK J T, et al. Domain Adaptation via Transfer Component Analysis [J/OL]. IEEE Transactions on Neural Networks, 2011, 22(2): 199-210. DOI: 10.1109/TNN.20 10.2091281.
[68]SMOLA A J, GRETTON A, BORGWARDT K. Maximum mean discrepancy[C]//13th international conference, ICONIP. 2006: 3-6.
[69]MURPHY R A, MONDRAGÓN E, MURPHY V A. Rule learning by rats[J]. Science, 2008, 319(5871): 1849-1851.
[70]CARUANA R. Multitask learning[M]. Springer, 1998.
[71]WIESE J, SAPONAS T S, BRUSH A B. Phoneprioception: enabling mobile phones to infer where they are kept[C]//Proceedings of the SIGCHI conference on human factors in computing systems. 2013: 2157-2166.
[72]YANG J, MUNGUIA-TAPIA E, GIBBS S. Efficient in-pocket detection with mobile phones [C]//Proceedings of the 2013 ACM conference on Pervasive and ubiquitous computing adjunct publication. 2013: 31-34.
[73]WINTER E. The shapley value[J]. Handbook of game theory with economic applications, 2002, 3: 2025-2054.
[74]ZHENG Y, LI G, LI Y, et al. Truth inference in crowdsourcing: Is the problem solved?[J]. Proceedings of the VLDB Endowment, 2017, 10(5): 541-552.