I am taking AI this semester and we have a semester project. We can choose just about anything. I was wondering if anyone has a creative idea that I might be able to do. Nothing that is so extensive that it cannot be finished in four weeks. Any help would be so appreciated!

Some background information: I am a graduate student in CS, but this is my first AI course. My research area is in the space of data mining and analytics. I am open to doing anything that seems interesting and creative.

  • 1
    $\begingroup$ Adding your background knowledge will help in the recommendations. $\endgroup$ – DuttaA Apr 1 at 19:07
  • 1
    $\begingroup$ I am not even sure this is on-topic here. I am voting to close it, as I don't think this is on topic. @KateCatelena, maybe visit our chat: chat.stackexchange.com/rooms/43371/the-singularity for a more interactive discussion. Maybe there you will be luckier! $\endgroup$ – nbro Apr 1 at 21:56
  • $\begingroup$ The question is not how to make an innovative project. It is, as stated, "What are some creative ideas for a semester project in an AI course?". This definitely on topic here. $\endgroup$ – John Doucette Apr 1 at 23:13
  • $\begingroup$ @JohnDoucette According to the on-topic page (which likely needs some refactoring), this website is not about "asking for a development tool or career path recommendation", and it doesn't mention that recommendations for projects are on-topic. Futhermore, this question will lead to broad answers. Maybe Kate can rephrase the question to make it more specific. $\endgroup$ – nbro Apr 2 at 8:23
  • 1
    $\begingroup$ I reopened b/c John makes a solid argument that this is relevant academically, and provided a good answer with several upvotes. My sense is the question is specific--qualifying the scope of such projects not to exceed 4 weeks. which might be difficult for a student to discern. $\endgroup$ – DukeZhou Apr 4 at 17:48

Welcome to AI.SE @Kate_Catelena!

I teach AI courses at the undergraduate level, and so have seen a lot of semester projects over the years. Here are some templates that often lead to exciting outcomes:

  1. Pick a new board or card game, and write a program to play it. Your course has probably covered Adversarial Search, and may also have covered Monte Carlo Tree Search, or self-play reinforcement learning approaches. These projects are often fun to mark and creative because they are easy enough to be well done, and yet there are always new, exciting, domains to apply these algorithms to. Some examples of past projects that I thought were neat were an AI to play the boardgame Tac (mostly A* Search), and an AI to play the card game Love Letter (mostly Counter-factual minimax regret, the algorithm used to solve poker).

  2. Pick a question that you would like to know the answer to, that could be addressed with machine learning. Then implement your own ML algorithm (decision tree learners are fairly easy), gather your own data, and show a result. Examples of interesting projects I've seen in the past are using ML to find out which of a number of factors most strongly influenced a students' subjective quality of sleep; and which items are most commonly purchased along with camping supplies (using association rule mining).

  3. Anything involving reinforcement learning. RL projects are always neat to see if accompanied by a visualization showing the learner's behavior at different stages. A strong past project involved a student simply replicating Sutton & Barto's Acrobot experiments with their own implementation of the SARSA-Lambda algorithm. Other things that might be neat include making a trainable "pet" that the user can influence, or solving games using self-play.

  4. Theoretical results might seem intimidating but are often more accessible than one might think, especially if your discrete math skills are strong. I have had many student projects where the student went away to look at theory papers in ML or Multiagent Systems, found a suggestion in the future work sections that wasn't a big result, but that was fairly easy to prove, and proved it. Sometimes these are even publishable.

  5. Replications. Go find an interesting AI paper (use scholar.google.com), and then see if you can do exactly what the authors suggest and if you get the same result or not. Then, if you have time, see if you can improve on the results. These are often most interesting when you find a paper written in a different field that uses AI. Often the authors of such papers know less about AI than you do, and so it can be fairly easy to improve on their results. I have had several students do projects like this to great effect.

Those categories are a bit vague, but remember: AI touches almost anything. Pick your favorite hobby, and see whether you can relate it to AI using one of the approaches above. Nothing makes a project stand out like one that applies AI to solve some real issues in an exciting domain. Good luck!

  • $\begingroup$ All of these seem group projects though otherwise it seems kinda impossible to finish in 4 weeks starting from grounds up. $\endgroup$ – DuttaA Apr 2 at 7:27
  • 1
    $\begingroup$ @DuttaA I find these are usually tractable for senior students to do in 3-4 weeks. It does depend on the student however! $\endgroup$ – John Doucette Apr 2 at 15:22
  1. Music Generation using GA/MA
  2. Open AI GYM Projects
  3. 2048 on RL and Search Algorithms
  4. Fixing a code

;) Enjoy


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.