New LLP-MM method improves learning from label proportions on high-dimensional data

Researchers have developed a new method called LLP via moment matching (LLP-MM) to address limitations in learning from label proportions (LLP) on high-dimensional data. Existing LLP methods often assume random data assembly, which is frequently not the case in real-world applications. LLP-MM, however, operates under a more realistic assumption where bags are sampled independently and instance labels are conditionally independent given the instances, leading to label counts that follow a Poisson multinomial distribution. This approach utilizes multi-order factorial moments as training objectives to better exploit the data's statistical structure, demonstrating effectiveness and computational efficiency in experiments. AI

IMPACT This new method could improve the accuracy of machine learning models trained with incomplete label information, particularly in scenarios where data is not randomly sampled.

RANK_REASON The item is an academic paper detailing a new machine learning method. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• alphaXiv
• arXiv
• CatalyzeX
• CORE Recommender
• DagsHub
• Gotit.pub
• Hugging Face
• IArxiv Recommender
• Learning from label proportions on high-dimensional data
• LLP-MM
• LLP via moment matching
• Poisson multinomial distribution
• ScienceCast

• paper
• other

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