Modeling high-order feature interactions is crucial for click-through rate (CTR) prediction, and traditional approaches often predefine a maximum interaction order and rely on exhaustive enumeration of feature combinations up to this predefined order. This framework heavily relies on prior domain knowledge to define interaction scope and entails high computational costs from enumeration. Conventional CTR models face a trade-off between improving representation through complex high-order feature interactions and reducing computational inefficiencies associated with these processes. To address this dual challenge, this study introduces the Kolmogorov-Arnold Represented Sparse Efficient Interaction Network (KarSein). Drawing inspiration from the learnable activation mechanism in the Kolmogorov-Arnold Network (KAN), KarSein leverages this mechanism to adaptively transform low-order basic features into high-order feature interactions, offering a novel approach to feature interaction modeling. KarSein extends the capabilities of KAN by introducing a more efficient architecture that significantly reduces computational costs while accommodating two-dimensional embedding vectors as feature inputs. Furthermore, it overcomes the limitation of KAN's its inability to spontaneously capture multiplicative relationships among features.Extensive experiments highlight the superiority of KarSein, demonstrating its ability to surpass not only the vanilla implementation of KAN in CTR predictio but also other baseline methods. Remarkably, KarSein achieves exceptional predictive accuracy while maintaining a highly compact parameter size and minimal computational overhead. As the first attempt to apply KAN in the CTR domain, this work introduces KarSein as a novel solution for modeling complex feature interactions, underscoring its transformative potential in advancing CTR prediction task.
View on arXiv@article{shi2025_2408.08713,
title={ Beyond KAN: Introducing KarSein for Adaptive High-Order Feature Interaction Modeling in CTR Prediction },
author={ Yunxiao Shi and Wujiang Xu and Haimin Zhang and Qiang Wu and Min Xu },
journal={arXiv preprint arXiv:2408.08713},
year={ 2025 }
}