tocarina/howto2/data/content

17 lines
2.3 KiB
Plaintext
Raw Normal View History

2022-02-24 11:53:43 +01:00
<section Content>
Your Report should include at least 4 Parts:
<list>
<e>Start by introducing your Topic. Why should your reader care enough to read your whole report?</e>
<e>Then you need to summarize what you have learned. This summary should be enough for a Student who missed your presentation to be able to understand a topic building on top of yours. For example a Student wanting to explain how backpropagation works in recurrent neural networks, should be able to understand how backpropagation works in "normal" neural networks.</e>
<e>If your assignment has explicit tasks (summarize 3 applications...) you need to include the results.</e>
<e>The next section is complicated to get right, but also the section we care the most about. We would like you to be critical, for the simple reason that we don't want to (only) grade how well you summarize your Topic. But we understand that it can be hard to criticize something as basic as a convolutional network. So you can extend this section to "everything that is not covered in your literature". For example you can compare your approach to alternatives (every topic should have alternative ways to achieve the same thing) and explain when to use which. You could also show some limitations and we appreciate any programming that falls under this section. Also in most weeks it can be justified to be critical. If you have problems with this, don't hesitate to ask your supervisors.</e>
<e>Finally you should summarize the rest of your report, this should not be long and is mostly used for your reader to remember your report quickly. Think of the conclusion in any paper.</e>
</list>
Afterwards you need to include all of your references (see below) and can attach less important parts (like output images of your programming assignments).
Please hand in programming assignments either over google colab, as a .zip folder attached to your email or as a link to a public git archive.
If these require any dependencies (python) or are written in any other language beside python please state this (Even through I suggest using python). You don't need to do this when using google colab, as dependencies are included and we can see when you install packages using !pip install. It can also be useful to include the output of your network and to make your network reproducible by seeding it.