Predicting Tomorrow’s Mood, Health, and Stress Level using Personalized Multitask Learning and Domain Adaptation

Natasha Jaques, Ognjen (Oggi) Rudovic, Sara Taylor, Akane Sano, Rosalind Picard
Proceedings of IJCAI 2017 Workshop on Artificial Intelligence in Affective Computing, PMLR 66:17-33, 2017.

Abstract

Predicting a person’s mood tomorrow, from data collected unobtrusively using wearable sensors and smartphones, could have a number of beneficial clinical applications; however, this prediction is an extremely challenging problem. Past approaches often lack the accurate and reliable performance necessary for real-world applications. We posit that this is due to the inability of traditional, one-size-fits-all machine learning models to account for individual differences. To overcome this, we treat predicting tomorrow’s mood for a single person as one task, or problem domain. We then adopt Multitask Learning (MTL) and Domain Adaptation (DA) approaches to learn a model which is customized for each person, while still being able to benefit from data across the population. Empirical results on real-world, continuous monitoring data show that the new personalized models — a MTL deep neural network, and a Gaussian Process with DA - both significantly outperform their generic counterparts, providing substantial performance enhancements in automatic prediction of continuous levels of tomorrow’s reported mood, stress,and physical health based on data through today.

Cite this Paper


BibTeX
@InProceedings{pmlr-v66-jaques17a, title = {Predicting Tomorrow’s Mood, Health, and Stress Level using Personalized Multitask Learning and Domain Adaptation}, author = {Jaques, Natasha and Rudovic, Ognjen (Oggi) and Taylor, Sara and Sano, Akane and Picard, Rosalind}, booktitle = {Proceedings of IJCAI 2017 Workshop on Artificial Intelligence in Affective Computing}, pages = {17--33}, year = {2017}, editor = {Lawrence, Neil and Reid, Mark}, volume = {66}, series = {Proceedings of Machine Learning Research}, month = {20 Aug}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v66/jaques17a/jaques17a.pdf}, url = {https://proceedings.mlr.press/v66/jaques17a.html}, abstract = {Predicting a person’s mood tomorrow, from data collected unobtrusively using wearable sensors and smartphones, could have a number of beneficial clinical applications; however, this prediction is an extremely challenging problem. Past approaches often lack the accurate and reliable performance necessary for real-world applications. We posit that this is due to the inability of traditional, one-size-fits-all machine learning models to account for individual differences. To overcome this, we treat predicting tomorrow’s mood for a single person as one task, or problem domain. We then adopt Multitask Learning (MTL) and Domain Adaptation (DA) approaches to learn a model which is customized for each person, while still being able to benefit from data across the population. Empirical results on real-world, continuous monitoring data show that the new personalized models — a MTL deep neural network, and a Gaussian Process with DA - both significantly outperform their generic counterparts, providing substantial performance enhancements in automatic prediction of continuous levels of tomorrow’s reported mood, stress,and physical health based on data through today.} }
Endnote
%0 Conference Paper %T Predicting Tomorrow’s Mood, Health, and Stress Level using Personalized Multitask Learning and Domain Adaptation %A Natasha Jaques %A Ognjen (Oggi) Rudovic %A Sara Taylor %A Akane Sano %A Rosalind Picard %B Proceedings of IJCAI 2017 Workshop on Artificial Intelligence in Affective Computing %C Proceedings of Machine Learning Research %D 2017 %E Neil Lawrence %E Mark Reid %F pmlr-v66-jaques17a %I PMLR %P 17--33 %U https://proceedings.mlr.press/v66/jaques17a.html %V 66 %X Predicting a person’s mood tomorrow, from data collected unobtrusively using wearable sensors and smartphones, could have a number of beneficial clinical applications; however, this prediction is an extremely challenging problem. Past approaches often lack the accurate and reliable performance necessary for real-world applications. We posit that this is due to the inability of traditional, one-size-fits-all machine learning models to account for individual differences. To overcome this, we treat predicting tomorrow’s mood for a single person as one task, or problem domain. We then adopt Multitask Learning (MTL) and Domain Adaptation (DA) approaches to learn a model which is customized for each person, while still being able to benefit from data across the population. Empirical results on real-world, continuous monitoring data show that the new personalized models — a MTL deep neural network, and a Gaussian Process with DA - both significantly outperform their generic counterparts, providing substantial performance enhancements in automatic prediction of continuous levels of tomorrow’s reported mood, stress,and physical health based on data through today.
APA
Jaques, N., Rudovic, O.(., Taylor, S., Sano, A. & Picard, R.. (2017). Predicting Tomorrow’s Mood, Health, and Stress Level using Personalized Multitask Learning and Domain Adaptation. Proceedings of IJCAI 2017 Workshop on Artificial Intelligence in Affective Computing, in Proceedings of Machine Learning Research 66:17-33 Available from https://proceedings.mlr.press/v66/jaques17a.html.

Related Material