Semi-deep learning with user feedback improving energy prediction models and improving user experience

16:19-16:31, January 27 @ 2BC

Talk/ Overview

We introduce a semi-supervised feedback labeling mechanism to improve the performance of deep learning based energy prediction models, where we used a GAN-based framework to help us automatically generate the labels of the feedback from our customers. Based on the GAN labeled feedback we either use them as a new dataset to create new energy prediction models or improve our existing models.

Talk/ Speakers

Jiufeng Shi

Data Scientist, Discovergy GmbH

AMLD / Global partners