mlsanityData to 14 Oct 2026

Learning Cascade Ranking as One Network

arXiv 2503.09492 ↗First posted 12 Mar 2025cs.IR cs.LGcited by 2

Not on the map. Why?

Abstract (arXiv)

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.

Key facts. Cited by 2 papers in the map (data through 14 Oct 2026). Builds on 28 of its 45 references that are in the map. First posted on arXiv 12 Mar 2025.

Builds on

28 of 45 references.

References · 28

  1. 2017Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networkscited by 4,897
  2. 2015Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classificationcited by 3,611
  3. 2017Multi-Task Learning Using Uncertainty to Weigh Losses for Scene Geometry and Semanticscited by 1,358
  4. 2017Deep Interest Network for Click-Through Rate Predictioncited by 544
  5. 2018End-to-End Multi-Task Learning with Attentioncited by 534
  6. 2020Just Pick a Sign: Optimizing Deep Multitask Models with Gradient Sign Dropoutcited by 201
  7. 2020Fast Differentiable Sorting and Rankingcited by 197
  8. 2018Entire Space Multi-Task Model: An Effective Approach for Estimating Post-Click Conversion Ratecited by 173
  9. 2020Gradient Vaccine: Investigating and Improving Multi-task Optimization in Massively Multilingual Modelscited by 169
  10. 2018Task-Agnostic Meta-Learning for Few-shot Learningcited by 151
  11. 2019Differentiable Ranks and Sorting using Optimal Transportcited by 113
  12. 2019Stochastic Optimization of Sorting Networks via Continuous Relaxationscited by 111
16 more
  1. 2020SoftSort: A Continuous Relaxation for the argsort Operatorcited by 66
  2. 2021NeuralNDCG: Direct Optimisation of a Ranking Metric via Differentiable Relaxation of Sortingcited by 41
  3. 2022On the Effectiveness of Sampled Softmax Loss for Item Recommendationcited by 39
  4. 2022Towards Understanding the Overfitting Phenomenon of Deep Click-Through Rate Prediction Modelscited by 33
  5. 2021Differentiable Sorting Networks for Scalable Sorting and Ranking Supervisioncited by 31
  6. 2020PiRank: Scalable Learning To Rank via Differentiable Sortingcited by 27
  7. 2023Fast, Differentiable and Sparse Top-k: a Convex Analysis Perspectivecited by 26
  8. 2022Monotonic Differentiable Sorting Networkscited by 25
  9. 2022Joint Optimization of Ranking and Calibration with Contextualized Hybrid Modelcited by 22
  10. 2023Tree based Progressive Regression Model for Watch-Time Prediction in Short-video Recommendationcited by 18
  11. 2022MBCT: Tree-Based Feature-Aware Binning for Individual Uncertainty Calibrationcited by 11
  12. 2024Full Stage Learning to Rank: A Unified Framework for Multi-Stage Systemscited by 7
  13. 2024RecFlow: An Industrial Full Flow Recommendation Datasetcited by 7
  14. 2023Adaptive Neural Ranking Framework: Toward Maximized Business Goal for Cascade Ranking Systemscited by 6
  15. 2022CROLoss: Towards a Customizable Loss for Retrieval Models in Recommender Systemscited by 4
  16. 2024XB-MAML: Learning Expandable Basis Parameters for Effective Meta-Learning with Wide Task Coveragecited by 3

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 1; 2026: 1.

2 papers.

  1. 2025Differentiable Fast Top-K Selection for Large-Scale Recommendationcited by 3
  2. 2026FunnelAL: Retrieve-then-Rank Active Learning for Single-Class Discoverycited by 0