Implicitly-Defined Neural Networks for Sequence Labeling
Abstract
We relax the causality assumption in formulating recurrent neural networks, so that the hidden states of the network are all coupled together. This goes beyond bidirectional RNN, which consists of two explicit recurrent networks concatenated together. The motivation behind doing this is to improve performance on long-range dependencies, and to improve stability (solution drift) in NLP tasks. We choose an implicit neural network architecture, show that it can be computed reasonably efficiently, and demonstrate it proof of-concept on the task of part-of-speech tagging.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 09, 2016
- Accession Number
- AD1033419
Entities
People
- Michaeel Kazi
Organizations
- Massachusetts Institute of Technology