New ICED model estimates density for tabular data without retraining

Researchers have developed ICED, a novel in-context, energy-based density estimator designed for tabular data. This transformer-based model is pre-trained on a synthetic dataset and can perform multiple unsupervised tasks, including anomaly detection, out-of-distribution detection, and data augmentation, without requiring per-dataset fitting or hyperparameter tuning. ICED achieves competitive performance across these tasks with a single, frozen model, eliminating the need for specialized pipelines and retraining. AI

IMPACT This model could streamline multiple unsupervised learning tasks on tabular data by providing a single, adaptable solution.

RANK_REASON The cluster describes a new research paper detailing a novel model for tabular data density estimation. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• alphaXiv
• arXiv
• CatalyzeX
• DagsHub
• Gotit.pub
• Hugging Face
• IArxiv
• ICED
• ScienceCast
• Transformer++

• paper
• model release
• other

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