You may choose following projects.
- Implement a recent research paper.
- You need to implement it in a different framework than available code. If available code is in keras, you can implement using pytorch or gluon mxnet.
- You need to repeat the experiments and note the different results between the paper and your output code
- You will write a report that summarizes paper and its contributions.
- In the same report, you will note the difficulties you encountered.
- You will present your paper and results.
Always check the paper with me. And inform me about your paper of choice via email.
-
Enter a Kaggle competition. Kaggle competition you enter should not be a getting started or example competition.
-
You can use interesting Kaggle Dataset or any other dataset. Check with me.
-
Introduction
- Team members
- What team members did for project
-
Problem Description
-
Dataset Description
- Where to download
- Any preprocessing
- problems encountered???
-
Neural Network Architecture used
- If any known architecture used reference it.
- if your own trial, explain a bit.
- you can use visualization of NN in this part too.
-
Training testing results
- some figures
-
Conclusion
I prefer brief reports to long reports.
- What is it?
- How to avoid it.
Using common sentence structure in your reports is not Plagiarism. You can easily take skeleton structure of common sentence and use it in your reports. A good example for common skeleton sentences in academic writing is in following Academic Phrase Bank web site .
An example from Academic Phrase bank site: "Referring to data in a table or chart"
| an overview of … | ||
|---|---|---|
| Table1 | shows | the breakdown of X according to … |
| Figure 1 | compares | |
| presents | ||
| ... |
- Online table creation in markdown. Markdown is markup language for jupyter notebooks.