H90-1053 productions were computed by an iterative procedure described below . The parser
C04-1091 be computed iteratively by an iterative procedure called local search . In each
H05-1043 ) . Relaxation labeling is an iterative procedure whose output is an assignment
E97-1024 factor for the cell entries . The iterative procedure has the following steps : 1 .
H91-1025 . This suggests the following iterative procedure for obtaining a good 2 : 1 .
D12-1045 overall benefit is found . This iterative procedure is the core of our joint coreference
D13-1016 ( 1 ) t The EM algorithm is an iterative procedure . In each iteration , it takes
A00-2033 CFGs attacks the problem by an iterative procedure for transforming indirect left
D12-1060 for OPTIM we can formulate the iterative procedure in the following way : fk ( xi
D11-1088 . The algorithm consists of an iterative procedure using the above three rules until
H90-1069 previous , related domain . An iterative procedure then revises the probability
D14-1210 operation , as opposed to the iterative procedure of EM , and therefore is much
D13-1111 constructs diverse lists via a greedy iterative procedure as follows . Let y1 be the model-best
E03-1026 Previous approaches use either iterative procedures to re-estimate alignment parameters
D15-1121 we still advocate the general iterative procedure sketched in Algo . 1 . Initialization
H01-1074 sliding window , which uses an iterative procedure to maximize the score of a segment
C04-1031 clue ma - trix . However , simple iterative procedures come very close to the reference
C92-2099 alternative parse trees nsing an iterative procedure analogous to the inside-outside
E87-1010 for each value , and then use an iterative procedure to repeatedly improve on these
C80-1070 + and it must be solved by an iterative procedure . Therefore , the uniquness of
hide detail