Deep learning training and analysis framework

  • Deep learning training and analysis framework
EEECS Summer Research Internships 2018

Proposed Project Title:

  • Deep learning training and analysis framework

Principal Supervisor(s):
  • John Bustard


Project Description:

MIT magazine[1][2] has recently published a top 10 list of breakthrough technologies they believe will make the most difference over the next 12 months, 3 of which are directly based on deep learning technology:

  1. AI for everybody (More accessible cloud based deep learning)
  2. Duelling neural networks (Artificial creativity)
  3. Babel-fish earbuds (realtime translation)

Two other areas are also very closely linked

  1. Sensing city (AI inferences from IoT city data)
  2. Genetic fortune telling (AI predictions from DNA)

The progress, impact and investment in this technology is unprecedented. While the core algorithms of deep learning have been developed by computer scientists and mathematicians, applying them effectively requires a very different skill set. The goal of this project is to create a framework to enable the teaching and research of deep learning models.

 

[1] https://www.technologyreview.com/magazine/2018/03/
[2] https://www.dezeen.com/2018/02/22/mit-technology-review-predicts-10-breakthrough-technologies/


Objectives:

The system to be developed should assist with the organisation of stored datasets and planned training of models using different deep learning systems. The project also involves the  development of procedures to ensure there is a good scientific approach to testing and analysing deep learning systems and that the process of developing and refining architectures and hyperparameters is captured so that we can learn from each system that is created.

Within both the makerspace and IoT lab we have a number of different PCs capable of training and deploying deep learning algorithms. Over the summer we are also developing a high bandwidth sensing framework that will be a testbed for many existing and novel deep learning algorithms. The project will include integrating with this system so that it can easily make use of the gathered data and algorithms can be easily deployed within that system.


Academic Requirements:

The scheme is open to all EEECS Undergraduates (apart from students on the BIT degree pathway and students who are due to graduate this summer)

This project is perfect for a student keen to learn about current deep learning frameworks and who wants to study how the best practicioners in the world develop their systems. Applicants should be willing to learn frameworks and develop software on linux (and debug systems issues with drivers etc.) and be interested in learning enough python, lua etc. to be able to inteface with the existing frameworks.


General Information:

Each internship will last between 6-8 weeks and will pay a weekly stipend of £250.

Start date: (Flexible based on student availability provided the project is complete before the start of term)

Duration:  8 (Weeks)

Location: Computer Science Building (IoT Lab)

Further information available at: http://www.qub.ac.uk/schools/eeecs/Research/


Contact details:

Supervisor Name: John Bustard
Address:

Queens University of Belfast
School of EEECS,
Computer Science Building,
18 Malone Road,
Belfast
BT9 5BN

Email: j.bustard@qub.ac.uk