On rpm inspired problems however we believe you cant

This preview shows page 6 - 8 out of 13 pages.

on RPM-inspired problems; however, we believe you can’t design an agent that does well on RPM-inspired problems without also accomplishing those learning goals. Is it possible to accomplish the learning goals without designing a successful agent? Yes, which brings us to the next section... Take Chances, Make Mistakes As you’re working on these projects, you may encounter a dilemma that students have reported encountering in the past. On the one hand, you might have an idea for a way of designing your agent that you believe would do pretty well, but may not be all that interesting. On the other hand, you might have an idea for a way of designing your agent that may not work at all, but would be more innovative and novel. You might hesitate to try out this latter method because there is some risk involved: it might not work, and because your grade is based on the number of problems your agent gets right, your performance might suffer because you tried to do something more interesting. If you find yourself in this dilemma, take the risk. Try out the approach you find more innovative, novel, or interesting. If the approach ends up being successful, great! If the approach does not end up being successful, however, write about it thoroughly in your project reflection. Describe the “safe” idea that you did not use, as well as the novel idea that you did use. Explore why you believed the novel idea had potential, and try and explain why you believe it was not successful: were you unable to implement it successfully, or was there an inherent problem in the method? How will this
experience inform your next project? How would you improve on this method in the future? If you take a novel approach to the project and that novel approach does not pan out, we will compensate for that in your grade. In line with the learning goals above, we want you to take risks and explore innovative or interesting approaches to these problems. If you take a risk and it doesn’t pay off, your grade won’t suffer; just make sure to explain the risk you took in detail in your project reflection, as well as why it didn’t pan out. In other words, demonstrate that you accomplished the learning goals of the project even though your agent did not perform that well. Project Difficulty and Authenticity As you go through these projects, you might realize as students have in the past that they are not easy. The projects in the course are very difficult. However, they are not difficult simply for the sake of being difficult: they are difficult because they address a real problem that the artificial intelligence community is facing right now. There exists ongoing research by professors, research scientists, and PhD students right now with the exact same goal as your project here. Former KBAI students are pursuing this research in ongoing Master’s theses. If the problem was easy, it wouldn’t be the subject of ongoing research -- it would have been solved long ago.

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture