Research map / Label-efficient and robust learning / Recognition and long-tail
Evaluating Recommender System Stability with Influence-Guided Fuzzing
David Shriver, Sebastian Elbaum, Matthew B. Dwyer, David S. Rosenblum
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 |
|---|---|---|---|
| guide fuzzing with influence models via AMR (rating influence) instead of random additions | User-User recommender · MovieLens 100k, 1 modification | TopOut instability ratio: 128x | Improved |
| use AMR instead of random additions | FunkSVD · MovieLens 100k, 10 modifications | TopOut instability ratio: over 40000x | Improved |
| use AMA (attribute influence) instead of random additions | LightFM · MovieLens 100k, 10 and 100 modifications | TopOut instability ratio: 1.3x and 3.2x | Improved |
| use AMA instead of random additions | User-User, Item-Item, FunkSVD (no attribute use) · MovieLens 100k | TopOut instability | No clear effect |
| use AMU, ALI, AMR instead of random additions | LightFM · MovieLens 100k | TopOut instability | No clear effect |
| use remove-type RMU and change-type CMU instead of random | Item-Item · MovieLens 100k | TopOut instability ratio: 6.3x and 13.4x | Improved |
| use Add-type influence heuristics instead of random additions | Item-Item · MovieLens 100k | TopOut instability | No clear effect |
| use APORTFOLIO round-robin of four heuristics instead of ARAND | User-User · MovieLens 100k, 100 modifications | mean TopOut instability: 0.031294 -> 0.257794 | Improved |
| use APORTFOLIO instead of ARAND | FunkSVD · MovieLens 100k, 100 modifications | mean TopOut instability: 0.053446 -> 0.768982 | Improved |
| use APORTFOLIO instead of ARAND | Item-Item · MovieLens 100k, 100 modifications | mean TopOut instability: 0.026681 -> 0.008653 | Hurt |
| use APORTFOLIO instead of ARAND | LightFM · MovieLens 100k, 100 modifications | mean TopOut instability: 0.010933 -> 0.017678 | Improved |
Closest papers by meaning
- Deconvolving Feedback Loops in Recommender Systems (NeurIPS 2016 · similarity 0.85)
- On Component Interactions in Two-Stage Recommender Systems (NeurIPS 2021 · similarity 0.85)
- Influential Exemplar Replay for Incremental Learning in Recommender Systems (AAAI 2024 · similarity 0.84)
- User-Creator Feature Polarization in Recommender Systems with Dual Influence (NeurIPS 2024 · similarity 0.84)
- CURE4Rec: A Benchmark for Recommendation Unlearning with Deeper Influence (NeurIPS 2024 · similarity 0.82)
- Would You Like Your Data to Be Trained? A User Controllable Recommendation Framework (AAAI 2024 · 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.
