Research map / Learning theory and algorithms / Sample complexity
Learning Structured Distributions From Untrusted Batches: Faster and Simpler
Sitan Chen, Jerry Li, Ankur Moitra
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 |
|---|---|---|---|
| Use LearnWithFilter with piecewise-constant post-processing for structured distributions | Learning from untrusted batches · Synthetic structured distributions, total variation | estimation error | Improved |
| Increase batch size k | LearnWithFilter and the oracle estimator · Synthetic arbitrary distributions, A_{ℓ/2} norm | estimation error | Improved |
| Increase batch size k | Naive empirical mean · Synthetic arbitrary distributions, A_{ℓ/2} norm | estimation error | No clear effect |
| Increase corruption fraction ε | Naive empirical mean · Synthetic arbitrary distributions, A_{ℓ/2} norm | estimation error | Hurt |
| Vary domain size n while keeping the batch count independent of n | LearnWithFilter · Synthetic arbitrary distributions, A_{ℓ/2} norm | estimation error | No clear effect |
| Use the structured estimator instead of the empirical mean of uncorrupted samples | Learning structured distributions · Synthetic structured distributions, total variation | estimation error | Improved |
Closest papers by meaning
- The Power of Iterative Filtering for Supervised Learning with (Heavy) Contamination (NeurIPS 2025 · similarity 0.85)
- Certified Machine Unlearning via Noisy Stochastic Gradient Descent (NeurIPS 2024 · similarity 0.83)
- The Utility and Complexity of In- and Out-of-Distribution Machine Unlearning (ICLR 2025 · similarity 0.83)
- Rewind-to-Delete: Certified Machine Unlearning for Nonconvex Functions (NeurIPS 2025 · similarity 0.83)
- Learning Juntas under Markov Random Fields (NeurIPS 2025 · similarity 0.83)
- Sample-Optimal Agnostic Boosting with Unlabeled Data (ICML 2025 · 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.
