Repeat Until Bored: A Pattern Selection Strategy
Paul Munro · Neural Information Processing Systems · 1991
An alternative to the typical technique of selecting training examples independently from a fixed distribution is formulated and analyzed, in which the current example is presented repeatedly until the error for that item is reduced to some criterion value, β; then, another item is randomly selected. The convergence time can be dramatically increased or decreased by this heuristic, depending on the task, and is very sensitive to the value of β.