Training a whole-book LSTM-based recognizer with an optimal training set.
Mohammad Reza SoheiliMohammad Reza YousefiEhsanollah KabirDidier StrickerPublished in: ICMV (2017)
Keyphrases
- training set
- test set
- training samples
- training examples
- supervised learning
- training data
- training patterns
- data sets
- book presents
- decision trees
- book covers
- dynamic programming
- classification accuracy
- recurrent neural networks
- hidden markov models
- early stopping
- training set size
- class distribution
- neural network
- database applications
- nearest neighbor
- active learning
- machine learning
- cross validation
- worst case
- error rate
- test images
- reinforcement learning
- researchers and practitioners
- feature space
- cutting edge
- handwriting recognition
- database
- computer based training
- support vector machine