Research map / Learning theory and algorithms / Prediction and decision losses
Fast Greedy MAP Inference for Determinantal Point Process to Improve Recommendation Diversity
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 incremental Cholesky updates for exact greedy DPP MAP inference instead of Lazy | DPP MAP inference · Synthetic DPP MAP inference | speedup: about 100 times faster than Lazy | Helped |
| Use incremental Cholesky updates for exact greedy DPP MAP inference instead of ApX | DPP MAP inference · Synthetic DPP MAP inference | speed | Helped |
| Use the exact FaX implementation instead of approximate ApX | DPP MAP inference · Synthetic DPP MAP inference | accuracy: ApX loses about 0.2% accuracy; FaX accuracy is the same as Lazy | Helped |
| Use DPP recommendations instead of MMR | Movie recommender · Online movie recommender A/B test | improvement of No. Titles Watched: 0.84% -> 1.33% | Helped |
| Use DPP recommendations instead of MMR | Movie recommender · Online movie recommender A/B test | improvement of Watch Minutes: 0.86% -> 1.52% | Helped |
| Use DPP rather than MMR, MSD, Entropy, or Cover for relevance-diversity ranking | Recommendation ranking · Netflix Prize and Million Song Dataset | relevance-diversity trade-off | Helped |
| Increase the relevance weight θ from low values toward a moderate value | DPP recommendation · Netflix Prize | MRR | Mixed |
| Increase the relevance weight θ | DPP recommendation · Netflix Prize | ILAD | Hurt |
| Increase the relevance weight θ | DPP recommendation · Netflix Prize | ILMD | Hurt |
| Use DPP rather than MMR or MSD for sliding-window diversification | Long-sequence recommendation · Netflix Prize and Million Song Dataset | relevance-diversity trade-off | Helped |
| Use DPP rather than Cover | Recommendation ranking · Netflix Prize and Million Song Dataset | relevance-diversity trade-off | Mixed |
Closest papers by meaning
- Lazy and Fast Greedy MAP Inference for Determinantal Point Process (NeurIPS 2022 · similarity 0.91)
- Faster Greedy MAP Inference for Determinantal Point Processes (ICML 2017 · similarity 0.90)
- A Tree-Based Method for Fast Repeated Sampling of Determinantal Point Processes (ICML 2019 · similarity 0.89)
- Maximizing Induced Cardinality Under a Determinantal Point Process (NeurIPS 2018 · similarity 0.89)
- Scalable Learning and MAP Inference for Nonsymmetric Determinantal Point Processes (ICLR 2021 · similarity 0.88)
- State-of-art minibatches via novel DPP kernels: discretization, wavelets, and rough objectives (NeurIPS 2026 · 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.