| H92-1035 | . We call this training method | 1-best training | . The SNN was originally trained |
| N12-1054 | P Akwk2 + 1 P p -- 1 min w 4.3 | 1-Best Training | For the final pass , we want |
| N12-1054 | margin performs better than other | 1-best training | algorithms . <title> Space Efficiencies |
| H92-1035 | which used the SNN produced by the | 1-best training | as an initial estimate ) , the |
| H92-1035 | diflicult . This is better than the | 1-best training | algorithm , which only uses the |
| H92-1035 | training is more effective than the | 1-best training | in the N-best rescoring paradigm |
| H92-1035 | l-layer SNN was trained using the | 1-best training | algorithm and the MSE criterion |
| H92-1035 | SEGMENTAL NEURAL NET TRAINING | 1-Best Training | In our original training algorithm |
