Researchers have developed a new method called Adam-Aware In-Run Data Shapley to accurately attribute data contributions in machine learning models that use the Adam optimizer. Existing methods, which rely on the linear structure of Stochastic Gradient Descent (SGD), are ineffective with Adam due to its complex dynamics. The proposed technique achieves near-perfect fidelity to ground-truth marginal contributions and maintains approximately 95% of standard training throughput. This new approach significantly outperforms SGD-based baselines in downstream data attribution tasks. AI
IMPACT Improves data attribution accuracy for modern ML training pipelines using Adam optimizer.
RANK_REASON The cluster contains an academic paper detailing a new method for machine learning optimization. [lever_c_demoted from research: ic=1 ai=1.0]
- Adam-Aware In-Run Data Shapley
- Adam optimizer
- arXiv
- Linearized Ghost Approximation
- Meng Ding
- Shapley value
- stochastic gradient descent
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
