Seminar 1283

2017/18: Deep Learning for Text Classification

Johann Mitlöhner

In the first seminar unit you will select one of the base papers as the topic for your own seminar work.

• It would be wise to read at least some of the abstracts before the first seminar unit, so you can make an informed choice.
• It would be even better to read all abstracts and at least 2-3 of your preferred papers since the papers will be assigned first come/first serve.

Your paper has to be written in LaTeX. Study the LaTeX Info for help. A short overview will be presented in the first unit.

1. Base paper selection -- 12.10.2017 17:00-20:00 D1.1.078
   upload five keywords by 13. 10. 2017
   After that:
   → read the base paper (if you have not done so already at this point..)
   → collect and read additional sources in order to get a good understanding of your topic
   → tutorials on deep learning and neural nets will help, such as deeplearning4j.org/neuralnet-overview
   → prepare 2-3 page proposal including 3 research questions and submit via learn
2. Consultation -- 16.11.2017 10:00-18:00
   we will talk individually about your proposal and select one of the research questions for further work. Email to set up time.
   After that:
   → prepare your SE paper based on your proposal
   → prepare your presentation
3. Presentation -- 14.12.2017 13:00-17:00 D5.1.004
   give us a 10 minute presentation about your work
   After that:
   → refine your SE paper
4. Submit final SE paper -- by 31. 1. 2018
   → switch on your spell checker
   → proofread your text
   → use Learn for submission of PDF file, email zip archive containing all your files

Additional material:

• Notes on writing your paper, including plagiarism and copyright issues
• Python notebook for computational introduction to neural nets

Other organisational details: VVZ