New PII detector achieves high recall with lower cost · arXiv research

Researchers have developed a new PII detection model that is both broad and efficient, addressing the limitations of existing detectors which often rely on complex architectures and limited synthetic datasets. This new model, trained on over a million examples from 77 public datasets and relabeled under a unified definition, achieves higher recall on key benchmarks like AI4Privacy and Nemotron-PII compared to larger baselines. Notably, it accomplishes this with significantly lower computational costs, offering 2-10x higher throughput and requiring substantially less GPU memory. AI

IMPACT This model could significantly reduce the cost and improve the effectiveness of PII detection in large-scale data processing.

RANK_REASON The cluster contains an academic paper detailing a new model and its evaluation on benchmarks. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CL →

• AI4Privacy
• arXiv
• Hugging Face
• Nemotron-PII
• Pii

• paper
• safety

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