| S15-2160 | possible Laplace ratio in every | feature adding | iteration . This is contrary |
| P98-2214 | improvements over the one-by-one | feature adding | algorithm with much smaller number |
| P98-2214 | advantage over the one-by-one | feature adding | algorithm , especially when the |
| P98-2214 | from the sequence of one-by-one | feature adding | in section 3.1 by the procedure |
| P98-2214 | Optimal " model of " One-by-one | Feature Adding | " method is the one selected |
| P98-2214 | feature selection by one-by-one | feature adding | considers all the non-active |
| W08-0302 | sentence . 3 Source-Side Context | Features Adding | features to the linear model |
| P98-2214 | timal " model of " One-by-one | Feature Adding | " method both in coverage and |
| P90-1024 | Also , the monotonicity of our " | feature adding | " ap - proach , as opposed to |
| P98-2214 | method and 800 for " One-by-one | Feature Adding | " method . Therefore , general-to |
| P98-1080 | Linuxpowered Pentium 90 , with | feature adding | threshold set to 4 ( which means |
| P98-2214 | Feature Selection by One-by-one | Feature Adding | The feature selection process |
