Examples of Student Projects
Three compulsory modules in our degrees ask students to undertake project work, providing the opportunity to complete a piece of extended work that puts into practice the techniques and theories they have learnt, and to investigate in greater depth an area of particular interest.
In Level 2, all students work in teams to complete a Group Project, and in Level 3, an Individual Project is done where each student works under the supervision of a member of academic staff. Students on the MEng programme also spend half of their time in their fourth year working on an Advanced Project.
The projects also introduce further skills relating to team management, oral presentations and report writing, essential graduate skills.
Level 2: Group Project
Every year tonnes of food go to waste in the UK. Much of the time, people simply throw food away that they will not use, yet would be willing to donate this unwanted food to those who need it.
Teams were required to develop an IT solution that operates on portable devices to promote food sharing in communities. The software includes functionality such as allowing geographically close people to post their unwanted food, and for others to search and find it, then arrange collection.
Whilst the project mandate is the same for each team, students are encouraged to use creative initiative to develop interesting variations of the system. The project is sponsored by IBM who award a prize to the team with the best project.
Level 3: Individual Project
Ben Hazelwood - Colours and Conflicts.
An important building block in scientific computing is simple stencil codes (these are loops over pixels, that might, for example, combine left, right, upper, lower neighbour and current pixel value into a new value). One of many scientific applications is to solve the fluid flow through a wind tunnel. The project looked at using shared memory nodes (found today even in commodity multicore computers), on which these codes are traditionally parallelised through colouring or (recursive) tiling. We evaluated traditional multithreading strategies and studied the arising assignment of tasks to threads and derived two efficient ways to parallelise stencil codes on regular grids.
The results were presented at International Workshops and a research paper was published. The work also led to discussions with Intel.
Level 4: Advanced Project
Leah Clark - An investigation into data collection techniques and deep learning to predict tennis matches.
Deep learning is a technique in which a computing system known as a deep neural network (DNN) is fed large quantities of data which it can then use to make decisions about other data. The aim of this project was to 'train' a DNN with past statistics of tennis matches so that it could predict the outcome of future matches. The project involved creating a data-extraction tool (to collect data from the web), building a data store, and testing the DNN which was shown to outperform other state-of-the-art methods.