Computer Laboratory

Course pages 2017–18

Deep learning for natural language processing

Principal lecturer: Dr Stephen Clark
Additional lecturer: (to be confirmed)
Taken by: MPhil ACS, Part III
Code: R228
Hours: 16 (16 1 hour lectures, 2 2-hour practicals)
Prerequisites: L101 Machine Learning for Language Processing, LE49 Probabilistic Machine Learning, Part II Machine Learning and Bayesian Inference, or an equivalent machine learning subject at undergraduate level. Note that L101 has a class limit of 16.

Teaching Style

Stephen Clark will give the introductory lectures. Some of the more specialised lectures will be delivered by DeepMind research scientists, including Felix Hill (a recent PhD graduate from the NLIP research group).

Aims and Objectives

Deep learning has become the dominant approach for a number of disciplines in Artificial Intelligence, including Natural Language Processing (NLP). The recent practical successes of Deep Learning, leading to improvements in fundamental NLP technologies such as machine translation, have created enormous interest in academia and industry.

By the end of the course, students will have an understanding of the fundamental techniques in deep learning and neural networks which enable the development of effective NLP applications. They will also have experience of using Google's Tensorflow software - one of the most popular development environments for building deep learning models - to solve NLP tasks.

Assessment

40% for the practical (based on a short report).

60% for one take-home exam at end of course.