| W04-2901 | discriminatively trained using maximum | mutual information estimation | . The language models are pruned |
| W98-1307 | Maximum-Likelihood , or Maximum | Mutual Information Estimation | -LSB- 18 -RSB- . The results |
| W06-1646 | by five iterations of maximum | mutual information estimation | ( MMI ) . Unlike other comparable |
| P04-1007 | parameters in the form of Maximum | Mutual Information Estimation | ( MMIE ) or Conditional Maximum |
| W09-2210 | usually referred to as Maximum | Mutual Information Estimation | ( MMIE ) and the second component |
