Federated Learning for Competitive Collaboration

People

Research areas

Description

Collaboration is a cornerstone of scientific research, and an essential element of engineering.  Successful collaborations and teams work best when data is shared openly, but in industry this is not always possible.  Many data assets contain proprietary information and the secrets to a company’s competitive advantage.  In the business sector data is rarely shared at all.  The recent development of Federated Learning by Google Research provides a means for overcoming this barrier, ensuring data remains private, but allowing researchers to collaborate and benefit from the combined results.  Researchers and professionals can compete and collaborate at the same time.

This project is ideal for an R&D, MComp or MCVML student with an interest in both machine learning and software engineering. You will develop a general federated learning module that is compatible with widely used platforms such as scikit-learn, keras and tensorflow, and test it using two different types of data sets from the engineering and business domains.  All programming will be done in python, and data sets will be provided.

This project is a collaboration with Tanjo.ai and involves regular (video conference) project meetings and discussions with Tanjo.ai staff in the USA.

This is a 24pc project.

 

Requirements

Completion of COMP6730, COMP6670 and COMP8430

Please double check my email when enquiring to ensure your message finds the right professor.

 

Background Literature

Keywords

machine learning, data science, software engineering, python

 

Updated:  1 June 2019/Responsible Officer:  Dean, CECS/Page Contact:  CECS Marketing