Interaction-Aware Structured Pruning for Click-Through Rate Prediction Models with Data-Adaptive Noise Gating
Electronics · 30.10.2026 · 15 (21) · 5031
DOI: 10.3390/electronics15215031 ↗Kurz gefasst
Fachartikel in der Zeitschrift Electronics, online erschienen am 30.10.2026. Verfasst von Beomhee Han und 1 weiteren Autor:innen. Die Arbeit nennt 39 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Research Foundation of Korea.
Abstract (Originalsprache)
Click-through rate (CTR) prediction models in real-time bidding (RTB) systems must remain accurate under strict latency budgets (about 20 ms for model inference). Explicit interaction architectures such as Deep & Cross Network V2 (DCN V2) combine large embedding tables with a cross-layer cost that grows quadratically with the input dimension. We propose Interaction-Aware Structured Pruning (IASP), which scores embedding dimensions by combining a Taylor-inspired importance score (magnitude and gradient) with each dimension’s participation in the cross network; a data-adaptive gate, whose threshold τ is a percentile of the interaction scores, restricts the synergy boost to dimensions with strong cross-layer participation. On the Avazu and Criteo datasets, IASP retains or improves the area under the ROC curve (AUC) at up to 99% embedding sparsity for DCN and DCN V2 and outperforms magnitude- and gradient-based criteria at 90–95% sparsity within a shared pruning pipeline. Physically removing the lowest-ranked dimensions (one dimension per field retained) reduces the memory footprint by about 16× and yields up to a 6.6× single-thread CPU speedup for DCN V2 when 200 candidates are ranked per request, whereas the speedup for single-candidate requests is marginal.
Förderung
- National Research Foundation of Korea
Zitieren
APA
Han, B., & Kim, C. O. (2026). Interaction-Aware Structured Pruning for Click-Through Rate Prediction Models with Data-Adaptive Noise Gating. Electronics, 15(21), 5031. https://doi.org/10.3390/electronics15215031
Vancouver
Han B, Kim CO. Interaction-Aware Structured Pruning for Click-Through Rate Prediction Models with Data-Adaptive Noise Gating. Electronics. 2026;15(21):5031. doi:10.3390/electronics15215031
BibTeX
@article{Han2026,
author = {Han, Beomhee and Kim, Chang Ouk},
title = {Interaction-Aware Structured Pruning for Click-Through Rate Prediction Models with Data-Adaptive Noise Gating},
journal = {Electronics},
year = {2026},
volume = {15},
number = {21},
pages = {5031},
doi = {10.3390/electronics15215031}
}
EUVIMED