Gradient Ascent as a concept transcends machine learning. It is the reverse of Gradient Descent, another common concept used in machine learning. Gradient Ascent (resp. Descent) is an iterative optimization algorithm used for finding a local maximum (resp. minimum) of a function.