Learning Cascade Ranking as One Network

Yunli Wang, ZhenZhang, Zhiqiang Wang, Zixuan Yang, Yu Li, Jian Yang, Shiyang Wen, Peng Jiang, Kun Gai

TL;DR: We propose LCRON, which improves cascade ranking systems by introducing an end-to-end training method that aligns with the overall system objective and enhances collaboration among stages.

Abstract: Cascade Ranking is a prevalent architecture in large-scale top-k selection systems like recommendation and advertising platforms. Traditional training methods focus on single-stage optimization, neglecting interactions between stages. Recent advances have introduced interaction-aware training paradigms, but still struggle to 1) align training objectives with the goal of the entire cascade ranking (i.e., end-to-end recall of ground-truth items) and 2) learn effective collaboration patterns for different stages. To address these challenges, we propose LCRON, which introduces a novel surrogate loss function derived from the lower bound probability that ground truth items are selected by cascade ranking, ensuring alignment with the overall objective of the system. According to the properties of the derived bound, we further design an auxiliary loss for each stage to drive the reduction of this bound, leading to a more robust and effective top-k selection. LCRON enables end-to-end training of the entire cascade ranking system as a unified network. Experimental results demonstrate that LCRON achieves significant improvement over existing methods on public benchmarks and industrial applications, addressing key limitations in cascade ranking training and significantly enhancing system performance.

Lay Summary:

Modern recommendation systems help users find what they truly care about—from movies to products—by selecting the most relevant items from a large pool.

We developed LCRON, a new method that improves how these systems rank and select top results. Unlike traditional approaches that treat each step separately, LCRON views the whole process as one system, learning how different stages work together to improve overall performance. A key benefit of LCRON is that it can be trained end-to-end, directly optimizing for the final goal: showing users the most relevant content.

Our experiments show that LCRON outperforms existing methods on both public benchmarks and industrial data. This means it has great potential to enhance user experience across services like e-commerce, streaming platforms, and online ads.

Application-Driven Machine Learning: This submission is on Application-Driven Machine Learning.

Primary Area: Applications->Everything Else

Keywords: Cascade Ranking Optimization, Surrogate Loss Optimization, Multi-Stage Coordination, End-to-End Training, Recommendation Systems

Submission Number: 10655