Reproduce “ScalarLens: Numerical Embeddings with Stable Coordinates and Contextual Responses for CTR 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
Heng Yao, Tianying Liu, Yulou Shu, Yong He, Chuan Yuan, Kaibin Qiu, Guowei Chen, Jiayu Zhao and 1 more
ScalarLens encodes numerical features by separating a scalar-determined coordinate from a context-dependent response. A monotone learned mesh preserves the coordinate, while bounded message passing adapts numerical embeddings; categorical embeddings and the CTR backbone are unchanged. Across three datasets and nine backbones, it ranks first in 25 of 27 settings, without establishing temporal, causal, or new-family generalization.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.