AdaStop framework optimizes DNN testing by managing labeling costs

Researchers have developed AdaStop, a novel framework designed to optimize the testing of deep neural networks (DNNs) by intelligently managing labeling costs. AdaStop addresses the challenge of determining an appropriate labeling budget by formulating testing as a cost-benefit decision process. It estimates the marginal fault discovery rate and halts labeling when this rate drops below a predefined threshold, thereby reducing unnecessary expenses while still identifying a significant portion of model faults. AI

IMPACT Optimizes DNN testing efficiency, potentially reducing costs for developers and researchers by minimizing unnecessary labeling.

RANK_REASON The item is a research paper published on arXiv detailing a new method for testing deep neural networks. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• AdaStop
• alphaXiv
• arXiv
• CatalyzeX Code Finder for Papers
• Connected Papers
• CORE Recommender
• cs.AI
• DagsHub
• deep neural network
• Gotit.pub
• Hugging Face
• IArxiv Recommender
• Influence Flower
• Litmaps
• machine learning
• ScienceCast
• scite Smart Citations

• paper
• infra

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →