Research Article

Mixed Script Identification Using Automated DNN Hyperparameter Optimization

Table 2

Corpus statistics show that (tokens and %age in corpus).

Corpus statistics
LanguageTokens%age in corpus

Eng10231122.7
Roman Urdu9723521.6
Hindi8756319.4
Bengali8567219.0
Saraiki7841217.4
Total tokens in corpus451193