@inproceedings{simbuild2024_2152,
	doi = {},
	url = {https://publications.ibpsa.org/conference/paper/?id=simbuild2024_2152},
	year = {2024},
	month = {May},
	publisher = {IBPSA-USA},
	author = {Aya Doma  and  Fatima Amara  and  Mohamed Ouf},
	title  = {A Parameter-based Transfer Learning Approach for Predicting Occupancy in Institutional Buildings},
	booktitle = {Proceedings of SimBuild Conference 2024},
	volume  = {11},
	isbn = {},
	address  = {Denver, Colorado},
	series  = {IBPSA-USA Building Simulation Conference},
	pages = {758--767},
	abstract = {Accurately representing buildings’ occupancy schedules has been crucial in understanding the energy flexibility of buildings as well as improving the allocation of building services and resources. However, the limited availability of data to model occupancy schedules for specific buildings has been challenging the development of an accurate prediction model. This study evaluates a parameter-based transfer learning scheme developed for time series prediction of occupancy schedules with limited data. The study focuses on an institutional building in Quebec, Canada as a case study. The results showed that transferring the knowledge of the pre-trained time series model has not only reduced the fitting computational power but also reduced the model’s training requirements while maintaining a mean absolute error of 3 occupants.},
	issn = {},
	Organisation = {IBPSA-USA},
	Editors = {}
}