Generation of Synthetic Training Data for an HMM-based Handwriting Recognition System.
Tamás VargaHorst BunkePublished in: ICDAR (2003)
Keyphrases
- training data
- writer independent
- training set
- hidden markov models
- handwritten words
- supervised learning
- learning algorithm
- character recognition
- test data
- decision trees
- training examples
- data sets
- classification accuracy
- labeled data
- training process
- domain knowledge
- class labels
- classification models
- feature values
- generation process
- recognition algorithm
- noisy data
- test set
- training samples
- real world
- recognition process
- semi supervised learning
- support vector machine
- generation method
- handwriting recognition
- handwritten characters
- feature extraction
- recognition engine
- real images are presented
- learned from training data