Deep Neural Networks for Natural Language Processing

Dnr:

SNIC 2016/5-45

Type:

SNAC Small

Principal Investigator:

Marco Kuhlmann

Affiliation:

Linköpings universitet

Start Date:

2016-09-21

End Date:

2018-10-01

Primary Classification:

10208: Språkteknologi (språkvetenskaplig databehandling)

Webpage:

Allocation

Abstract

The goal of this project is to develop, implement, and evaluate new algorithms for natural language processing (NLP) based on deep neural networks. Modern network architectures such as the Long Short-Term Memory (LSTM) require efficient computation to process data at the scale that is now standardly being used in NLP applications. To meet this demand, the project will use state-of-the-art deep learning libraries with transparent support for GPU-based computation, such as Theano and TensorFlow.