gradient stability
Description: Gradient stability describes the reliability or bounded behavior of gradients during neural-network training, particularly in the design and analysis of activation functions and estimators.
Related topic: Neural Networks and Applications (11%)
OpenAlex ID: keywords/gradient-stability
