Sensitivity Sampling

Sensitivity sampling is a data subsampling technique aiming to create smaller, representative datasets for efficient machine learning. Current research focuses on improving sampling bounds and algorithms, particularly for high-dimensional data and specific loss functions, often incorporating techniques like k-means clustering and $\ell_p$ norm augmentation to enhance performance. This approach offers significant potential for improving the efficiency and scalability of machine learning models across various applications, from accelerating MRI reconstruction to ensuring fairness in predictive models. The development of tighter theoretical bounds and more efficient algorithms continues to be a major focus.