Byte-based Language Identification with Deep Convolutional Networks
This work addresses the problem of discriminating between similar languages for NLP applications, but it is incremental as it applies an existing method (ResNet) to a new data type (byte representations).
The authors tackled language identification for similar languages using a byte-based deep residual network, achieving accuracies of 84.88% on subtask A, 68.80% on subtask B1, and 69.80% on subtask B2.
We report on our system for the shared task on discriminating between similar languages (DSL 2016). The system uses only byte representations in a deep residual network (ResNet). The system, named ResIdent, is trained only on the data released with the task (closed training). We obtain 84.88% accuracy on subtask A, 68.80% accuracy on subtask B1, and 69.80% accuracy on subtask B2. A large difference in accuracy on development data can be observed with relatively minor changes in our network's architecture and hyperparameters. We therefore expect fine-tuning of these parameters to yield higher accuracies.