Pruning sparse non-negative matrix n-gram language models.

Joris Pelemans, Noam Shazeer, Ciprian Chelba

Browse the full Interspeech paper archive.