In this work we introduce an incremental learning framework for Click-Through-Rate (CTR) prediction and demonstrate its effectiveness for Taboola's massive-scale recommendation service. Our approach enables rapid capture of emerging trends through warm-starting from previously deployed models and fine tuning on "fresh" data only. Past knowledge is maintained via a teacher-student paradigm, where the teacher acts as a distillation technique, mitigating the catastrophic forgetting phenomenon. Our incremental learning framework enables significantly faster training and deployment cycles (x12 speedup). We demonstrate a consistent Revenue Per Mille (RPM) lift over multiple traffic segments and a significant CTR increase on newly introduced items.
@article{arxiv.2209.00458,
title = {An Incremental Learning framework for Large-scale CTR Prediction},
author = {Petros Katsileros and Nikiforos Mandilaras and Dimitrios Mallis and Vassilis Pitsikalis and Stavros Theodorakis and Gil Chamiel},
journal= {arXiv preprint arXiv:2209.00458},
year = {2022}
}
Comments
To be published in the Sixteenth ACM Conference on Recommender Systems (RecSys 22), Seattle, WA, USA