Research map / Bandits and online learning / Online convex optimization and games
Competitive Caching with Machine Learned Advice
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 Predictive Marker instead of LRU with PLECO predictions | Predictive Marker · BK, k=10 | competitive ratio: 1.280 -> 1.266 | Helped |
| Use Predictive Marker instead of LRU with PLECO predictions | Predictive Marker · Citi, k=100 | competitive ratio: 1.859 -> 1.810 | Helped |
| Use Predictive Marker instead of classical Marker | Predictive Marker · BK | competitive ratio: 1.310 -> 1.266 | Helped |
| Blindly trust the oracle (Blind Oracle) instead of Predictive Marker | Blind Oracle · BK with PLECO | competitive ratio: 1.266 -> 2.049 | Hurt |
| Blind Oracle instead of Predictive Marker on Citi | Blind Oracle · Citi with PLECO | competitive ratio: 1.810 -> 2.023 | Hurt |
| Increase lognormal oracle error sigma in synthetic predictions (ablation) | Predictive Marker vs LRU and Marker · BK synthetic, k=10 | ratio of evictions to optimum | Mixed |
| Increase oracle error sigma for the Blind Oracle (ablation) | Blind Oracle · BK synthetic | ratio of evictions to optimum | Hurt |
| Change cache size k (ablation) | all methods · BK and Citi PLECO | method ranking | No clear effect |
| Predictive Marker vs next best method LRU | Predictive Marker · BK and Citi PLECO | competitive ratio: 2.5% better | Helped |
Closest papers by meaning
- Robustifying Learning-Augmented Caching Efficiently without Compromising 1-Consistency (NeurIPS 2025 · similarity 0.90)
- Algorithms for Caching and MTS with reduced number of predictions (ICLR 2024 · similarity 0.88)
- Randomization Boosts KV Caching, Learning Balances Query Load: A Joint Perspective (ICLR 2026 · similarity 0.86)
- Parameterized Complexity of Caching in Networks (AAAI 2025 · similarity 0.85)
- DropKV: Decoupling Residual-Output Perturbation for Near-Optimal KV-Cache Eviction (NeurIPS 2026 · similarity 0.84)
- Learning-Augmented Algorithms with Explicit Predictors (NeurIPS 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.