Researchers have developed a new framework for stress-testing outlier detection methods used in subjective image quality assessment. This framework employs adversarial attacks to identify ratings that maximize discrepancies in Mean Opinion Score (MOS) estimates. The study demonstrates significant differences in the worst-case performance of various outlier detection methods under these adversarial conditions and proposes new, low-complexity methods that exhibit strong empirical worst-case performance. AI
IMPACT Introduces novel methods for robust data validation in image quality assessment, potentially improving the reliability of AI models trained on such data.
RANK_REASON The item is an academic paper detailing a new methodology and experimental results. [lever_c_demoted from research: ic=1 ai=0.7]
- alphaXiv
- arXiv
- CatalyzeX
- Connected Papers
- CORE Recommender
- DagsHub
- Dietmar Saupe
- Gotit.pub
- Hugging Face
- Influence Flower
- Litmaps
- ScienceCast
- scite Smart Citations
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →