Morphological Inflection Generation with Hard Monotonic Attention
arXiv:1611.01487
Abstract
We present a neural model for morphological inflection generation which employs a hard attention mechanism, inspired by the nearly-monotonic alignment commonly found between the characters in a word and the characters in its inflection. We evaluate the model on three previously studied morphological inflection generation datasets and show that it provides state of the art results in various setups compared to previous neural and non-neural approaches. Finally we present an analysis of the continuous representations learned by both the hard and soft attention \cite{bahdanauCB14} models for the task, shedding some light on the features such models extract.
Accepted as a long paper in ACL 2017
References in corpus (7)
- ADADELTA: An Adaptive Learning Rate Method
- Sequence Transduction with Recurrent Neural Networks
- Neural Machine Translation in Linear Time
- Online and Linear-Time Attention by Enforcing Monotonic Alignments
- Segmental Recurrent Neural Networks for End-to-end Speech Recognition
- A Neural Transducer
- Online Segment to Segment Neural Transduction
Cited by in corpus (5)
- Exploring Neural Transducers for End-to-End Speech Recognition
- Monotonic Chunkwise Attention
- An Exploration of Neural Sequence-to-Sequence Architectures for Automatic Post-Editing
- Linguistically inspired morphological inflection with a sequence to sequence model
- The Role of Interpretable Patterns in Deep Learning for Morphology