New active learning method uses Fermat distance for better data selection

Researchers have introduced PC-Fermat, a novel active learning method designed to improve data selection for labeling, particularly in low-budget scenarios. Unlike traditional methods that rely on Euclidean distance, PC-Fermat utilizes a Fermat distance to better reflect the data's underlying distribution. This approach constructs a weighted Euclidean neighborhood graph and employs a truncated shortest-path algorithm to compute the Fermat coverage graph, ensuring accuracy in distance calculations. Experiments on eight classification datasets show that PC-Fermat outperforms existing active learning methods, especially when labeling budgets are very limited. AI

IMPACT This method could reduce data labeling costs in machine learning projects, especially those with limited budgets.

RANK_REASON This is a research paper published on arXiv detailing a new method for active learning. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• alphaXiv
• arXiv
• arXivLabs
• CatalyzeX Code Finder for Papers
• Connected Papers
• CORE Recommender
• DagsHub
• Euclidean distance
• Fermat distance
• Gotit.pub
• Hugging Face
• IArxiv Recommender
• Influence Flower
• Litmaps
• PC-Fermat
• ProbCover
• ScienceCast
• scite Smart Citations

• paper
• other

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