COMP4691/8691 Optimisation

Seminar Overview (COMP8691 ONLY)

Students will work individually or in groups of up to 3 to research an optimisation topic that is not covered in the course material. Topics may include:

• A particularly interesting application of optimisation
• An optimisation algorithm or solving technique

The groups first need to propose a topic that will then be approved (or not) by the course lecturers. Be sure to do this as soon as you can so there is enough time to find another topic if your first choice is taken or inappropriate.

Once the topic is approved, the group will then further research the topic and put together a 15 minute video presentation on it.

The video recordings will be due in the last week of the semester. They will be uploaded for the other students to view in this week, in place of regular lectures.

Group Formation and Topic Proposal

We will use ed discussions to coordinate groups and topics. If you are looking for a group, create a new post on ed discussions, under the seminar-groups folder. From experience you are more likely to get a response if you also suggest a topic / set of topics that interest you.

Once you have established a group and chosen an interesting topic, post it on ed discussions under the seminar-topics folder. Your post should include the title of your seminar, the student members of the group, what you propose to cover in 15 minutes, and links to one or two papers or books that you have found to support your topic. The teaching staff will either then approve the topic or recommend changes.

Structure / Style

This is pretty flexible, you could decide to present a regular seminar with slides, a camera recording of you talking and working on paper, or you may decide to present more of a tutorial on a particular technique with coding examples. Or some combination of the above. If you have more wacky ideas, feel free to suggest them to us.

The video should be delivered at the level where another student taking this course can understand the content.

It is up to you how you split the work up in your teams. All of you can present, or just one, with the others contributing more to the research / video production etc aspects.

There should be at least a few seconds of recorded video of your faces at the start of the video giving an introduction, so that we can identify the speakers / team.

You should make use of at least 3 resources (papers or books) when developing your talk, and these should be clearly stated at the end of the talk. Figures / results or other content used in slides should be clearly referenced throughout.

If your topic is focused primarily on one paper (e.g., a specialised application with interesting results), then be sure to look at the related literature for a comparison of alternate approaches / results.

If you would like to do your own implementation / experiments relevant to the topic and present that as part of the video, this is also welcome.

You will be marked on a range of aspects such as on your delivery and how engaging the video is, the breadth and depth of research, the relevance to the course, the accuracy and the technical difficulty.

Production

One tool that might be useful is OBS (https://obsproject.com/) for simultaneously recording your screen and video from cameras. We don’t require super-high production quality, but your video should be clear and keep the audience engaged. There is a whole range of free video-editing software as well that we can suggest, or if you have some experience please give some recommendations to your fellow classmates.

Submission

Only one member of each group needs to submit your video. Submissions will be handled via email or wattle (to be defined). The file size limit is 200MB.

Example Topics

Technique-related topics:

• Conjugate gradient method
• Branch and price
• Sat solving for CSP / Lazy-clause generation
• No goods
• Accelerated methods
• Bilevel optimisation
• Stochastic programming
• Chance constraints
• Robust optimisation
• Multi-objective optimisation techniques
• Semidefinite programming
• Alternating direction method of multipliers
• Various metaheuristics
• Constraint-based scheduling
• Lazy-constraints in MIP
• Logic-based Benders decomposition

Some industries / fields to look for interesting optimisation applications:

• Mining
  • Mine planning
  • Mineral exploration
• Chemical processing
• Airlines
  • Tail assignment
  • Hub location
• Shipping
• Finance
  • Arbitrage
  • Options pricing
  • Portfolio optimisation
  • Asset-liability management
• Energy
  • Network design
  • Fault location
  • Optimal operation
• Machine learning
• Robotics
• Control
• Transport
  • Green logistics
  • Traffic signal optimisation
  • Bike share networks
  • Ride-sharing (dial-a-ride)
• Video games
  • Multi-agent path finding
• Environment
  • Habitat restoration
  • Habitat linking
  • Pest control
• Manufacturing
  • Assembly line design
  • Job shop scheduling
  • Order picking