Research map / Label-efficient and robust learning / Recognition and long-tail
FinalMLP: An Enhanced Two-Stream MLP Model for CTR Prediction
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use FinalMLP instead of EDCN | CTR prediction · Criteo | AUC: 81.47 -> 81.49 | Helped |
| Use FinalMLP instead of DualMLP | CTR prediction · Avazu | AUC: 76.57 -> 76.66 | Helped |
| Use FinalMLP instead of DualMLP | CTR prediction · MovieLens | AUC: 96.98 -> 97.20 | Helped |
| Use FinalMLP instead of EDCN | CTR prediction · Frappe | AUC: 98.50 -> 98.61 | Helped |
| Add stream-specific feature selection using context-aware feature gating (ablation) | FinalMLP · Four open benchmark datasets | AUC | Helped |
| Replace summation fusion with bilinear fusion (ablation) | FinalMLP · Four open benchmark datasets | AUC | Helped |
| Replace concatenation fusion with bilinear fusion (ablation) | FinalMLP · Four open benchmark datasets | AUC | Helped |
| Replace element-wise product fusion with bilinear fusion (ablation) | FinalMLP · Four open benchmark datasets | AUC | Helped |
| Use 10 rather than 1 bilinear-fusion head (ablation) | FinalMLP · Avazu | AUC: 0.7649 -> 0.7669 | Helped |
| Use 10 rather than 1 bilinear-fusion head (ablation) | FinalMLP · MovieLens | AUC: 0.9691 -> 0.9724 | Helped |
| Use 50 rather than 10 bilinear-fusion heads (ablation) | FinalMLP · Avazu | AUC: 0.7669 -> 0.7657 | Hurt |
| Use 50 rather than 10 bilinear-fusion heads (ablation) | FinalMLP · MovieLens | AUC: 0.9724 -> 0.9703 | Hurt |
Closest papers by meaning
- From Feature Interaction to Feature Generation: A Generative Paradigm of CTR Prediction Models (ICML 2025 · similarity 0.89)
- Addressing Cold-Start Problem in Click-Through Rate Prediction via Supervised Diffusion Modeling (AAAI 2025 · similarity 0.86)
- AT4CTR: Auxiliary Match Tasks for Enhancing Click-Through Rate Prediction (AAAI 2024 · similarity 0.85)
- Generalize for Future: Slow and Fast Trajectory Learning for CTR Prediction (AAAI 2024 · similarity 0.84)
- Learning Cascade Ranking as One Network (ICML 2025 · similarity 0.83)
- Beyond Graph Convolution: Multimodal Recommendation with Topology-aware MLPs (AAAI 2025 · similarity 0.83)
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.