📷 Paper + Code Links 🌐 Project page: https://t.co/OkqKPlel9X 📷 Paper: https://t.co/l1is7TzDaz 💻 GitHub repo: https://t.co/95Vsm9RCI9 Grateful to my co-authors and looking forward to sharing more at AAAI 2026!

📷 How does DaSH perform? ► Digit-Five: +26.2% accuracy ► DomainNet: +10.8% accuracy ✅ Requires far fewer exploration steps ✅ Robust even when group assignments are noisy

💡 We formalize dataset selection as its own problem and introduce DaSH (Dataset Selection via Hierarchies), a hierarchical Bayesian method that models dataset-level utility and group-level utility (e.g., source/institution) Instead of treating all datasets independently, DaSH:

🔍 The problem Most data selection methods: • operate at the instance level • assume all datasets are equally useful • break down under domain shift But in practice, data comes in datasets, grouped by institutions, repositories, or collections. Selecting the right

Excited to share our #AAAI2026 Oral paper Hierarchical Dataset Selection for High-Quality Data Sharing! 🤔 How do you decide which datasets to train on when data comes from many noisy, heterogeneous sources? 🧵