Nonparametric Teaching

Nonparametric teaching focuses on efficiently training machine learning models, particularly those representing complex functions without relying on predefined parametric forms, by strategically selecting training examples. Current research explores algorithms for both single and multiple learners, often employing overparameterized neural networks as the learner and focusing on optimizing example selection for faster convergence. This approach offers significant potential for improving the efficiency and scalability of training complex models across various applications, including image processing and other domains requiring the learning of non-parametric functions.