Research map / Learning theory and algorithms / Prediction and decision losses
Learning-Augmented Moment Estimation on Time-Decay Models
Soham Deepak Nagawanshi, Shalini Panthangi, Chen Wang, David Woodruff, Samson Zhou
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 |
|---|---|---|---|
| add heavy-hitter oracle augmentation to AMS for l2 norm estimation | AMS sliding-window · CAIDA l2 norm | estimate accuracy vs ground truth: AMS ratio 1.25-2.3 -> AMSA within 1.2 | Improved |
| add heavy-hitter oracle augmentation to SS for l3 norm (Count-Sketch, LLM, LSTM oracles) | SS sliding-window · CAIDA l3 norm | estimate accuracy | Improved |
| add oracle augmentation to SS for l3 norm | SS sliding-window · AOL l3 norm | estimate accuracy | Mixed |
| add oracle augmentation to SS under distribution shift | SS sliding-window · Synthetic l3 norm with distribution shift | estimate accuracy | Improved |
| use true sliding-window SSA instead of rescaling the full-window estimate (ablation) | SSA · Synthetic l3 norm with distribution shift | estimate accuracy | Improved |
| add oracle augmentation to SS for (k,p)-cascaded norm | SS sliding-window · Synthetic (k,p)-cascaded norm | estimate accuracy | Improved |
| add oracle augmentation to SS for (k,p)-cascaded norm | SS sliding-window · Synthetic (k=5,p=5) cascaded norm | RAM (MB): 74.63 -> 68.86 | Improved |
| add oracle augmentation to SS for (k,p)-cascaded norm | SS sliding-window · Synthetic (k=5,p=5) cascaded norm | running time (s): 63.5s -> 40.3s | Improved |
Closest papers by meaning
- Learning-Augmented Data Stream Algorithms (ICLR 2020 · similarity 0.89)
- Learning-Augmented Frequent Directions (ICLR 2025 · similarity 0.88)
- The Heavy Hitter Oracle: Enhancing Frequency Estimation in Skewed Data Streams (NeurIPS 2026 · similarity 0.86)
- Near-Optimal Streaming Heavy-Tailed Statistical Estimation with Clipped SGD (NeurIPS 2024 · similarity 0.84)
- Continual Release Moment Estimation with Differential Privacy (NeurIPS 2025 · similarity 0.83)
- Instance-Conditional Timescales of Decay for Non-Stationary Learning (AAAI 2024 · 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.
