Learning Personalized Thermal Preferences via Bayesian Active Learning with Unimodality Constraints

Thermal preferences vary from person to person and may change over time. The main objective of this paper is to sequentially pose intelligent queries to occupants in order to optimally learn the indoor air temperature values which maximize their satisfaction. Our central hypothesis is that an occupant's preference relation over indoor air temperature can be described using a scalar function of these temperatures, which we call the "occupant's thermal utility function". Information about an occupant's preference over these temperatures is available to us through their response to thermal preference queries : "prefer warmer," "prefer cooler" and "satisfied" which we interpret as statements about the derivative of their utility function, i.e. the utility function is "increasing", "decreasing" and "constant" respectively. We model this hidden utility function using a Gaussian process prior with built-in unimodality constraint, i.e., the utility function has a unique maximum, and we train this model using Bayesian inference. This permits an expected improvement based selection of next preference query to pose to the occupant, which takes into account both exploration (sampling from areas of high uncertainty) and exploitation (sampling from areas which are likely to offer an improvement over current best observation). We use this framework to sequentially design experiments and illustrate its benefits by showing that it requires drastically fewer observations to learn the maximally preferred temperature values as compared to other methods. This framework is an important step towards the development of intelligent HVAC systems which would be able to respond to occupants' personalized thermal comfort needs. In order to encourage the use of our PE framework and ensure reproducibility in results, we publish an implementation of our work named GPPrefElicit as an open-source package in Python.

[1]  H. B. Gunay,et al.  Modelling and analysis of unsolicited temperature setpoint change requests in office buildings , 2018 .

[2]  Gail Brager,et al.  Expectations of indoor climate control , 1996 .

[3]  Joyce Kim,et al.  Personal comfort models: Predicting individuals' thermal preference using occupant heating and cooling behavior and machine learning , 2018 .

[4]  Jasper Snoek,et al.  Practical Bayesian Optimization of Machine Learning Algorithms , 2012, NIPS.

[5]  Oliver Sawodny,et al.  Feature selection and Gaussian Process regression for personalized thermal comfort prediction , 2019, Building and Environment.

[6]  Athanasios Tzempelikos,et al.  A Bayesian approach for probabilistic classification and inference of occupant thermal preferences in office buildings , 2017 .

[7]  Peter Green,et al.  Markov chain Monte Carlo in Practice , 1996 .

[8]  Aki Vehtari,et al.  CORRECTING BOUNDARY OVER-EXPLORATION DEFICIENCIES IN BAYESIAN OPTIMIZATION WITH VIRTUAL DERIVATIVE SIGN OBSERVATIONS , 2017, 2018 IEEE 28th International Workshop on Machine Learning for Signal Processing (MLSP).

[9]  Yong Chan Kim,et al.  Simulation-based optimization of an integrated daylighting and HVAC system using the design of experiments method , 2016 .

[10]  R. Dear,et al.  Thermal adaptation in the built environment: a literature review , 1998 .

[11]  Iason Konstantzos,et al.  Inferring personalized visual satisfaction profiles in daylit offices from comparative preferences using a Bayesian approach , 2018, Building and Environment.

[12]  J. Berger Statistical Decision Theory and Bayesian Analysis , 1988 .

[13]  Ilias Bilionis,et al.  Model predictive control under forecast uncertainty for optimal operation of buildings with integrated solar systems , 2018, Solar Energy.

[14]  Alberto Cerpa,et al.  Thermovote: participatory sensing for efficient building HVAC conditioning , 2012, BuildSys@SenSys.

[15]  Burcin Becerik-Gerber,et al.  Infrared thermography of human face for monitoring thermoregulation performance and estimating personal thermal comfort , 2016 .

[16]  Liu Yang,et al.  Thermal comfort and building energy consumption implications - A review , 2014 .

[17]  Daphne Koller,et al.  Making Rational Decisions Using Adaptive Utility Elicitation , 2000, AAAI/IAAI.

[18]  Srinivasan Keshav,et al.  SPOT: a smart personalized office thermal control system , 2013, e-Energy '13.

[19]  Gail Brager,et al.  Indoor Environmental Quality ( IEQ ) Title Personal comfort models — A new paradigm in thermal comfort for occupant-centric environmental control Permalink , 2018 .

[20]  J. F. Nicol,et al.  The validity of ISO-PMV for predicting comfort votes in every-day thermal environments , 2002 .

[21]  J. Neumann,et al.  Theory of games and economic behavior , 1945, 100 Years of Math Milestones.

[22]  Ian Beausoleil-Morrison,et al.  Development and implementation of a thermostat learning algorithm , 2018 .

[23]  Iakovos Michailidis,et al.  Occupancy-based demand response and thermal comfort optimization in microgrids with renewable energy sources and energy storage , 2016 .

[24]  Kwok Wai Tham,et al.  Room air temperature affects occupants' physiology, perceptions and mental alertness , 2010 .

[25]  村上 昌史,et al.  Field experiments on energy consumption and thermal comfort in the office environment controlled by occupants' requirements from PC terminal , 2009 .

[26]  P. O. Fanger,et al.  Thermal comfort: analysis and applications in environmental engineering, , 1972 .

[27]  Chuan Zhou,et al.  Collaborative Dynamic Sparse Topic Regression with User Profile Evolution for Item Recommendation , 2017, AAAI.

[28]  Farrokh Jazizadeh,et al.  Personalized thermal comfort inference using RGB video images for distributed HVAC control , 2018, Applied Energy.

[29]  Lai Jiang,et al.  Modelling personal thermal sensations using C-Support Vector Classification (C-SVC) algorithm , 2016 .

[30]  Mohammad. Rasul,et al.  Thermal-comfort analysis and simulation for various low-energy cooling-technologies applied to an office building in a subtropical climate , 2008 .

[31]  A. Kennedy,et al.  Hybrid Monte Carlo , 1988 .

[32]  David B. Dunson,et al.  Bayesian Data Analysis , 2010 .

[33]  Scott Sanner,et al.  Real-time Multiattribute Bayesian Preference Elicitation with Pairwise Comparison Queries , 2010, AISTATS.

[34]  S. K. Smith,et al.  US Department of Energy Chernobyl Database , 1992 .

[35]  Andrew Gordon Wilson,et al.  Generalised Wishart Processes , 2010, UAI.

[36]  Nicolas Morel,et al.  A personalized measure of thermal comfort for building controls , 2011 .

[37]  E. Vázquez,et al.  Convergence properties of the expected improvement algorithm with fixed mean and covariance functions , 2007, 0712.3744.

[38]  A. Rosenfeld,et al.  Estimates of Improved Productivity and Health from Better Indoor Environments , 1997 .

[39]  Neil D. Lawrence,et al.  Preferential Bayesian Optimization , 2017, ICML.

[40]  Shengbo Guo,et al.  Bayesian recommender systems : models and algorithms , 2011 .

[41]  Iakovos Michailidis,et al.  Intelligent energy and thermal comfort management in grid-connected microgrids with heterogeneous occupancy schedule , 2015 .

[42]  Zoubin Ghahramani,et al.  Collaborative Gaussian Processes for Preference Learning , 2012, NIPS.

[43]  Lior Rokach,et al.  Recommender Systems: Introduction and Challenges , 2015, Recommender Systems Handbook.

[44]  Adrian Leaman,et al.  Productivity in buildings: the ‘killer’ variables , 1999 .

[45]  Tsuyoshi Murata,et al.  {m , 1934, ACML.

[46]  Ming Jin,et al.  Longitudinal Assessment of Thermal and Perceived Air Quality Acceptability in Relation to Temperature, Humidity, and CO2 Exposure in Singapore , 2017 .

[47]  Rajib Rana,et al.  Feasibility analysis of using humidex as an indoor thermal comfort predictor , 2013 .

[48]  S. Karjalainen,et al.  Thermal comfort and gender: a literature review. , 2012, Indoor air.

[49]  Junta Nakano,et al.  Differences in perception of indoor environment between Japanese and non-Japanese workers , 2002 .

[50]  Aki Vehtari,et al.  Gaussian processes with monotonicity information , 2010, AISTATS.

[51]  S. C.,et al.  A NOTE ON THE GAMMA DISTRIBUTION , 1958 .

[52]  Aki Vehtari,et al.  Bayesian Optimization of Unimodal Functions , 2017 .

[53]  Jonathan W. Pillow,et al.  Exploiting gradients and Hessians in Bayesian optimization and Bayesian quadrature. , 2017, 1704.00060.

[54]  Burcin Becerik-Gerber,et al.  Human-Building Interaction Framework for Personalized Thermal Comfort-Driven Systems in Office Buildings , 2014, J. Comput. Civ. Eng..

[55]  CARLOS A. GOMEZ-URIBE,et al.  The Netflix Recommender System , 2015, ACM Trans. Manag. Inf. Syst..

[56]  Radford M. Neal Pattern Recognition and Machine Learning , 2007, Technometrics.

[57]  Ian Dewancker,et al.  Sequential Preference-Based Optimization , 2018, ArXiv.

[58]  Eyke Hüllermeier,et al.  Preference-based Online Learning with Dueling Bandits: A Survey , 2018, J. Mach. Learn. Res..

[59]  Eyke Hüllermeier,et al.  Preference Learning , 2005, Künstliche Intell..

[60]  Eyke Hllermeier,et al.  Preference Learning , 2010 .

[61]  Burcin Becerik-Gerber,et al.  Energy savings from temperature setpoints and deadband: Quantifying the influence of building and system properties on savings , 2016 .

[62]  Burcin Becerik-Gerber,et al.  An online learning approach for quantifying personalized thermal comfort via adaptive stochastic modeling , 2015 .

[63]  Andrew Gordon Wilson,et al.  Scaling Gaussian Process Regression with Derivatives , 2018, NeurIPS.

[64]  M Frontczak,et al.  Quantitative relationships between occupant satisfaction and satisfaction aspects of indoor environmental quality and building design. , 2012, Indoor air.

[65]  Greg Linden,et al.  Two Decades of Recommender Systems at Amazon.com , 2017, IEEE Internet Computing.

[66]  Burcin Becerik-Gerber,et al.  User-led decentralized thermal comfort driven HVAC operations for improved efficiency in office buildings , 2014 .

[67]  Andreas Wagner,et al.  Thermal comfort and workplace occupant satisfaction—Results of field studies in German low energy office buildings , 2007 .

[68]  Madhavi Indraganti,et al.  Effect of age, gender, economic group and tenure on thermal comfort: A field study in residential buildings in hot and dry climate with seasonal variations , 2010 .

[69]  Alan E. Gelfand,et al.  On smoothness properties of spatial processes , 2003 .

[70]  Scott Sanner,et al.  Gaussian Process Preference Elicitation , 2010, NIPS.

[71]  Athanasios Tzempelikos,et al.  Inference of thermal preference profiles for personalized thermal environments with actual building occupants , 2019, Building and Environment.

[72]  Richard de Dear,et al.  Individual difference in thermal comfort: A literature review , 2018, Building and Environment.

[73]  Matthias Poloczek,et al.  Bayesian Optimization with Gradients , 2017, NIPS.

[74]  Srinivasan Keshav,et al.  Optimal Personal Comfort Management Using SPOT+ , 2013, BuildSys@SenSys.

[75]  H. Rijal,et al.  Thermal comfort in offices in India: Behavioral adaptation and the effect of age and gender , 2015 .

[76]  P Wargocki,et al.  Perceived air quality, sick building syndrome (SBS) symptoms and productivity in an office with two different pollution loads. , 1999, Indoor air.

[77]  Joseph A. Paradiso,et al.  Personalized HVAC control system , 2010, 2010 Internet of Things (IOT).

[78]  Alexis Boukouvalas,et al.  GPflow: A Gaussian Process Library using TensorFlow , 2016, J. Mach. Learn. Res..

[79]  Xin Liu,et al.  Modeling Users' Dynamic Preference for Personalized Recommendation , 2015, IJCAI.

[80]  Weiwei Liu,et al.  A neural network evaluation model for individual thermal comfort , 2007 .

[81]  D P Wyon,et al.  The effects of indoor air quality on performance and productivity. , 2004, Indoor air.