Title :
Symbol graph based discriminative training and rescoring for improved math symbol recognition
Author :
Luo, Zhen Xuan ; Shi, Yu ; Soong, Frank K.
Author_Institution :
Microsoft Res. Asia, Beijing
fDate :
March 31 2008-April 4 2008
Abstract :
In the symbol recognition stage of online handwritten math expression recognition, the one-pass dynamic programming algorithm can produce high-quality symbol graphs in addition of the best recognized hypotheses. In this paper, we exploit the rich hypotheses embedded in a symbol graph to discriminatively train the exponential weights of different model likelihoods and the insertion penalty. The training is investigated in two different criteria: maximum mutual information (MMI) and minimum symbol error (MSE). After discriminative training, trigram-based graph rescoring is performed in a post-processing stage. Experimental results finally show a 97% symbol accuracy on a test set of 2,574 written expressions with 43,300 symbols, a significant improvement of symbol accuracy obtained.
Keywords :
dynamic programming; graph theory; handwriting recognition; symbol manipulation; discriminative training; high-quality symbol graphs; improved math symbol recognition; insertion penalty; maximum mutual information; minimum symbol error; model likelihoods; one-pass dynamic programming algorithm; online handwritten math expression recognition; trigram-based graph rescoring; Asia; Computer science; Decoding; Dynamic programming; Handwriting recognition; Heuristic algorithms; Hidden Markov models; Mutual information; Speech recognition; Testing; Handwritten math formula recognition; discriminative training; graph rescoring; symbol graph; symbol recognition;
Conference_Titel :
Acoustics, Speech and Signal Processing, 2008. ICASSP 2008. IEEE International Conference on
Conference_Location :
Las Vegas, NV
Print_ISBN :
978-1-4244-1483-3
Electronic_ISBN :
1520-6149
DOI :
10.1109/ICASSP.2008.4518019