Research map / Probabilistic and scientific modeling / Molecules and proteins
DropoutNet: Addressing Cold Start in Recommender Systems
Maksims Volkovs, Guangwei Yu, Tomi Poutanen
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 |
|---|---|---|---|
| increase DNN depth from 1 to 3 layers (ablation) | DN-WMF · RecSys item cold start | recall@100: 0.234 -> 0.265 | Improved |
| increase DNN depth from 1 to 3 layers (ablation) | DN-WMF · RecSys user cold start | recall@100: 0.211 -> 0.231 | Improved |
| increase DNN depth from 1 to 3 layers (ablation) | DN-WMF · RecSys warm start | recall@100: 0.421 -> 0.412 | No clear effect |
| apply input preference dropout (rate 0 -> 0.1) (ablation) | DropoutNet on CiteULike · CiteULike cold start | recall@100: over 60% improvement | Improved |
| raise dropout rate up to 0.7 (ablation) | DropoutNet on CiteULike · CiteULike warm start | recall@100: decreases by less than 1% | No clear effect |
| raise dropout rate beyond 0.7 (ablation) | DropoutNet on CiteULike · CiteULike warm start | recall@100 | Hurt |
| raise dropout rate toward 1 (ablation) | DropoutNet on CiteULike · CiteULike cold start | recall@100 | Improved |
| wrap CDL with DropoutNet (DN-CDL) | CDL · CiteULike cold start | recall@100: 0.573 -> 0.629 | Improved |
| wrap CDL with DropoutNet (DN-CDL) | CDL · CiteULike warm start | recall@100: 0.603 -> 0.598 | No clear effect |
| use DN-WMF vs best baseline DeepMusic | DeepMusic · CiteULike cold start | recall@100: 0.601 -> 0.636 | Improved |
| feed preference information as input vs content-only (DeepMusic) on warm start | DeepMusic · CiteULike warm start | recall@100: 0.371 -> 0.593 | Improved |
| use tanh instead of ReLU or sigmoid activations (ablation) | DN-WMF · RecSys | recall@100 | Improved |
Closest papers by meaning
- Temporally and Distributionally Robust Optimization for Cold-Start Recommendation (AAAI 2024 · similarity 0.87)
- Infinite Recommendation Networks: A Data-Centric Approach (NeurIPS 2022 · similarity 0.87)
- Low-Rank Linear Cold-Start Recommendation from Social Data (AAAI 2017 · similarity 0.87)
- Feature-Structure Adaptive Completion Graph Neural Network for Cold-start Recommendation (AAAI 2025 · similarity 0.86)
- Actions Speak Louder than Words: Trillion-Parameter Sequential Transducers for Generative Recommendations (ICML 2024 · similarity 0.85)
- Wukong: Towards a Scaling Law for Large-Scale Recommendation (ICML 2024 · similarity 0.84)
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.
