Exportar Publicação

Labiadh, M., Obrecht, C., Ferreira da Silva, C. & Ghodous, P. (2019). On the suitability of Data Selection for Cross-building Knowledge Transfer. In The 17th International Conference on High Performance Computing Simulation (HPCS 2019), The 3rd Special Session on High Performance Services Computing and Internet Technologies (SerCo 2019). Dublin: IEEE.

M. Labiadh et al.,  "On the suitability of Data Selection for Cross-building Knowledge Transfer", in The 17th Int. Conf. on High Performance Computing Simulation (HPCS 2019), The 3rd Special Session on High Performance Services Computing and Internet Technologies (SerCo 2019), Dublin, IEEE, 2019

@inproceedings{labiadh2019_1734837676521,
	author = "Labiadh, M. and Obrecht, C. and Ferreira da Silva, C. and Ghodous, P.",
	title = "On the suitability of Data Selection for Cross-building Knowledge Transfer",
	booktitle = "The 17th International Conference on High Performance Computing Simulation (HPCS 2019), The 3rd Special Session on High Performance Services Computing and Internet Technologies (SerCo 2019)",
	year = "2019",
	editor = "",
	volume = "",
	number = "",
	series = "",
	doi = "10.1109/HPCS48598.2019.9188132",
	publisher = "IEEE",
	address = "Dublin",
	organization = "Bellatreche, L., Benouaret, K., and Hung, P.",
	url = "http://hpcs2019.cisedu.info/2-conference/special-sessions/session05-serco"
}

TY  - CPAPER
TI  - On the suitability of Data Selection for Cross-building Knowledge Transfer
T2  - The 17th International Conference on High Performance Computing Simulation (HPCS 2019), The 3rd Special Session on High Performance Services Computing and Internet Technologies (SerCo 2019)
AU  - Labiadh, M.
AU  - Obrecht, C.
AU  - Ferreira da Silva, C.
AU  - Ghodous, P.
PY  - 2019
DO  - 10.1109/HPCS48598.2019.9188132
CY  - Dublin
UR  - http://hpcs2019.cisedu.info/2-conference/special-sessions/session05-serco
AB  - Supervised deep learning has achieved remarkable success in various applications. Such advances were mainly attributed to the rise of computational powers and the amounts of training data made available. Therefore, accurate large-scale data collection services are often needed. One representative data is retrieved, it becomes possible to train the supervised machine learning predictor. However, a model trained on existing data, which generally comes from multiple datasets, might generalize poorly on the unseen target data. This problem is referred to as a domain shift. In this paper, we explore the suitability of data selection to tackle the domain shift challenge in the context of domain generalization. We perform our experimental study on the use case of building energy consumption prediction. Experimental results suggest that minimal building description is capable of improving cross-building generalization performances when used to select data.
ER  -