Reproduce “CAN: Feature Co-Action for Click-Through Rate Prediction” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Weijie Bian, Kailun Wu, Lejian Ren, Qi Pi, Yujing Zhang, Can Xiao, Xiang-Rong Sheng, Yong-Nan Zhu and 7 more
CAN addresses click-through-rate prediction by replacing separate learned representations for feature pairs with a reusable small network: one feature supplies its parameters, and the other supplies its input. On Amazon Books, CAN reaches an AUC of 0.7935 ± 0.0007 versus 0.7891 ± 0.0007 for Cartesian product, using 3.3× rather than 17.0× the same DNN baseline parameters. It is slightly lower on Taobao and Avazu.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.