This paper investigates the role of Deep Neural Networks (DNNs) in feature interaction recommendation models, addressing a debate on their ability to capture complex interactions. The research proposes a new perspective focusing on how DNNs affect the dimensional robustness of representations. Experiments with parallel and stacked DNNs show they can effectively prevent embedding dimensional collapse, with theoretical analysis revealing the underlying mechanisms. AI
Summary written by gemini-2.5-flash-lite from 3 sources. How we write summaries →
IMPACT Provides a theoretical and empirical understanding of DNNs' effectiveness in recommendation systems, potentially guiding future model design.
RANK_REASON This is a research paper published on arXiv.