Research map / Image generation / Fast diffusion sampling
iFusion: Integrating Dynamic Interest Streams via Diffusion Model for Click-Through Rate Prediction
Ziheng Ni, Congcong Liu, Yuying Chen, Zhiwei Fang, Changping Peng, Zhangang Lin, Ching Law, Jingping Shao
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 |
|---|---|---|---|
| add disentangled classifier-free guidance (DCFG) instead of no guidance (ablation) | iFusion · Industrial CTR dataset | AUC: 0.7607 -> 0.7685 | Improved |
| add DCFG instead of standard CFG (ablation) | iFusion · Industrial CTR dataset | AUC: 0.7663 -> 0.7685 | Improved |
| add MARN autoregressive denoising instead of NAR-MLP (ablation) | iFusion · Industrial CTR dataset | AUC: 0.7644 -> 0.7685 | Improved |
| add MARN instead of NAR attention network (ablation) | iFusion · Industrial CTR dataset | AUC: 0.7650 -> 0.7685 | Improved |
| use AR attention network instead of MARN (ablation) | iFusion · Industrial CTR dataset | AUC: 0.7685 -> 0.7689 | No clear effect |
| add consistency loss (ablation) | iFusion · Industrial CTR dataset | AUC: 0.7686 -> 0.7685 | No clear effect |
| add consistency loss (ablation) | iFusion · Industrial CTR dataset | inference speed: 12.9 -> 16.2 batches/sec | Improved |
| increase sampling steps under consistency training (ablation) | iFusion · Industrial CTR dataset | AUC | Hurt |
| set core-interest guidance gamma_cp to 2.0 and gamma_tf to 1.0 (ablation) | iFusion · Industrial CTR dataset | AUC: 0.7685 | Improved |
| replace best baseline MTGR with iFusion | iFusion · Amazon CTR dataset | AUC: 0.8440 -> 0.8512 | Improved |
| replace best baseline MTGR with iFusion | iFusion · Industrial CTR dataset | AUC: 0.7648 -> 0.7685 | Improved |
| deploy iFusion online versus production base | iFusion · Online A/B test | CTR: +2.44% | Improved |
Closest papers by meaning
- Addressing Cold-Start Problem in Click-Through Rate Prediction via Supervised Diffusion Modeling (AAAI 2025 · similarity 0.89)
- On Efficiency-Effectiveness Trade-off of Diffusion-based Recommenders (NeurIPS 2025 · similarity 0.88)
- From Feature Interaction to Feature Generation: A Generative Paradigm of CTR Prediction Models (ICML 2025 · similarity 0.87)
- Fading to Grow: Growing Preference Ratios via Preference Fading Discrete Diffusion for Recommendation (NeurIPS 2025 · similarity 0.87)
- Learning Multiple User Distributions for Recommendation via Guided Conditional Diffusion (AAAI 2025 · similarity 0.86)
- Plug-In Diffusion Model for Sequential Recommendation (AAAI 2024 · similarity 0.86)
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.
