MODE: Multi-Objective Adaptive Coreset Selection
This work addresses data efficiency for machine learning practitioners, but it appears incremental as it builds on existing coreset methods with adaptive improvements.
The paper tackles the problem of coreset selection for efficient training by introducing MODE, a framework that dynamically adapts selection strategies across training phases, achieving competitive accuracy with reduced memory requirements.
We present Mode(Multi-Objective adaptive Data Efficiency), a framework that dynamically combines coreset selection strategies based on their evolving contribution to model performance. Unlike static methods, \mode adapts selection criteria to training phases: emphasizing class balance early, diversity during representation learning, and uncertainty at convergence. We show that MODE achieves (1-1/e)-approximation with O(n \log n) complexity and demonstrates competitive accuracy while providing interpretable insights into data utility evolution. Experiments show \mode reduces memory requirements