TY - GEN
T1 - Optimizing the number of states for HMM-based on-line handwritten whiteboard recognition
AU - Geiger, Jürgen
AU - Schenk, Joachim
AU - Wallhoff, Frank
AU - Rigoll, Gerhard
PY - 2010
Y1 - 2010
N2 - In this paper, we present a novel way to determine the number of states in Hidden-Markov-Models for on-line handwriting recognition. This method extends the Bakis length modeling method which has succesfully been applied to off-line handwriting recognition. We propose a modification to the Bakis method and present a technique to improve the topology with a small number of iterations. Furthermore, we investigate the influence of state tying. In an experimental section, we show that our improved system outperforms a system with Bakis length modeling by 1.5 % relative and with fixed length modeling by 5.1 % relative on the IAM-On-DB-t1 benchmark.
AB - In this paper, we present a novel way to determine the number of states in Hidden-Markov-Models for on-line handwriting recognition. This method extends the Bakis length modeling method which has succesfully been applied to off-line handwriting recognition. We propose a modification to the Bakis method and present a technique to improve the topology with a small number of iterations. Furthermore, we investigate the influence of state tying. In an experimental section, we show that our improved system outperforms a system with Bakis length modeling by 1.5 % relative and with fixed length modeling by 5.1 % relative on the IAM-On-DB-t1 benchmark.
UR - http://www.scopus.com/inward/record.url?scp=79951713402&partnerID=8YFLogxK
U2 - 10.1109/ICFHR.2010.23
DO - 10.1109/ICFHR.2010.23
M3 - Conference contribution
AN - SCOPUS:79951713402
SN - 9780769542218
T3 - Proceedings - 12th International Conference on Frontiers in Handwriting Recognition, ICFHR 2010
SP - 107
EP - 112
BT - Proceedings - 12th International Conference on Frontiers in Handwriting Recognition, ICFHR 2010
T2 - 12th International Conference on Frontiers in Handwriting Recognition, ICFHR 2010
Y2 - 16 November 2010 through 18 November 2010
ER -