Fig 5 - uploaded by Jason Xiaotian Dou
Content may be subject to copyright.
Source publication
We review coreset optimization frameworks as a gateway to data-efficient machine learning. Coreset selection aims to select a subset of the most informative training samples which can achieve comparable performance as the full training sets. Mainstream approaches to coreset selection include submodular optimization, proxy models, gradient-based met...
Context in source publication
Context 1
... coreset optimization frameworks focus on the math and optimization modeling framework and treat low-level computation between CPU, RAM, and hard disk drive as a black box. However memories present a hierarchy structure as Figure 5. How should we re-image the coreset framework taking into account the memory hierarchy? ...