1 paper · 1 filter
Jon Kleinberg, Charlotte Peale, Omer Reingold
We investigate the learning task of language generation in the limit, but shift focus from the traditional time-of-last-mistake metric of a generator's success to a new notion of "…