Researchers have developed a new algorithm called Adversarial Contamination-resistant Iterative Hard Thresholding (AC-IHT) designed to handle high-dimensional regression problems affected by data contamination. This two-stage, nonconvex algorithm iteratively refines coefficient and contamination vectors to achieve near-optimal estimation. The AC-IHT algorithm is also signal-adaptive, capable of attaining sharper estimation rates and more accurate support recovery under specific signal conditions, while providing a theoretical foundation for asymptotic inference. AI
IMPACT Introduces a novel algorithm for robust statistical analysis in machine learning, potentially improving model reliability in the presence of noisy data.
RANK_REASON The cluster contains a research paper detailing a new algorithm for statistical machine learning.
- AC-IHT
- alphaXiv
- arXiv
- CatalyzeX Code Finder for Papers
- Connected Papers
- CORE Recommender
- DagsHub
- Gotit.pub
- Hugging Face
- Influence Flower
- Litmaps
- ScienceCast
- scite Smart Citations
- Adversarial Contamination-resistant Iterative Hard Thresholding
- stat.ML
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →