Conceptual
Browse Instruct Curate

Deep Learning Classification of Aviation Operational Records

Applying sequence-oriented neural networks to automatically sort free-text aviation operational records into categories such as Commercial, Military, and Private. A comparison of bidirectional LSTM, LSTM, convolutional, and simple recurrent architectures on the same labelled dataset shows how the ability to model long-range and bidirectional context, together with class balance, determines accuracy on domain-specific text classification.