Research map / Learning theory and algorithms / Prediction and decision losses
Delayed Feedback Modeling with Influence Functions
Chenlu Ding, Jiancan Wu, Yancheng Yuan, Cunchun Li, Xiang Wang, Dingxian Wang, Frank Yang, Andrew Rabinovich
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| apply influence-function parameter update for delayed conversions to the Vanilla model | Vanilla MLP · Criteo offline | AUC: 0.8353 -> 0.8411 | Improved |
| apply influence-function update | Vanilla DCNV2 · Criteo offline | AUC: 0.8370 -> 0.8410 | Improved |
| apply influence-function update vs FSIW (best baseline) | MLP · Criteo offline | Log loss: 0.4020 -> 0.3958 | Improved |
| use IF-DFM instead of ULC at larger gap c=14 | MLP · Criteo offline c=14 | AUC: 0.8345 -> 0.8420 | Improved |
| longer time gap c=14 for ULC and DFM | ULC/DFM · Criteo offline | AUC: Vanilla 0.8369 vs ULC 0.8345, DFM 0.8351 | Hurt |
| use nnDF with global information | nnDF · Criteo offline | AUC: 0.8353 (Vanilla) -> 0.6887 | Hurt |
| add newly arrived data update (delta theta add) to delay-only update (ablation) | IF-DFM online · Criteo online | AUC, PRAUC: +0.36% AUC, +0.47% PRAUC | Improved |
| use influence function update instead of retraining | IF-DFM · Criteo offline c=10 | running time (s): 1351.4 -> 14.8 | Improved |
| use ULC auxiliary label correction model | ULC · Criteo offline c=10 | running time (s): 1351.4 -> 2467.4 | Hurt |
Closest papers by meaning
- Online Conversion Rate Prediction via Multi-Interval Screening and Synthesizing under Delayed Feedback (AAAI 2024 · similarity 0.90)
- Capturing Delayed Feedback in Conversion Rate Prediction via Elapsed-Time Sampling (AAAI 2021 · similarity 0.89)
- Online Learning of Delayed Choices (NeurIPS 2024 · similarity 0.83)
- Towards Robust Influence Functions with Flat Validation Minima (ICML 2025 · similarity 0.82)
- Percentile Risk-Constrained Budget Pacing for Guaranteed Display Advertising in Online Optimization (AAAI 2024 · similarity 0.82)
- Entire-Space Variational Information Exploitation for Post-Click Conversion Rate Prediction (AAAI 2025 · similarity 0.82)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
