Presentation and Blog Post Guidelines

Shadi Albarqouni

AI Young Investigator Group Leader at Helmholtz AI | TUM Junior Fellow at TU Munich

05.11.2020

Presentation Guidelines

  1. **First Slide**: Please have the title of the paper, authors (<3), conference proceedings, or journal, and the name of the presenter.
``` Example:

The Future of Digital Health with Federated Learning

Rieke et al., Nature Digital Medicine, 2020

Presenter: Firstname Surname


---
1. <span class="fragment " >
    <span style="color:orange">**Introduction**:</span> In the first few slides, you need to introduce the subject to the audience. A brief **background** (big picture) and a few **related works** (more concise) would help you to position your paper in the big picture. *It is quite important to talk about the key conclusions at the very beginning*.  The **rationale** for the paper, i.e. *why you did the work?*, has to be addressed by the end of the Intro. slides. 
</span>

---
2. <span class="fragment " >
    <span style="color:orange">**Methodology**:</span> You need to explain the method in details, if possible. Start with an overview of the **framework** (e.g. flow chart); input, output, and core components, before you dive deeper into the **key contributions**; e.g. design architecture, objective functions, ...etc. Details that might distract the audience can be moved to the backup slides. In short, explain *how did you do it?*
</span>
---

3. <span class="fragment " >
    <span style="color:orange">**Experiments and Results**:</span> You need to explain the **experimental designs**; datasets, evaluation metrics, and training setup, and the **rationale behind them**, before you show the **key results**.  The figures should be clearly labeled, e.g. explain the figures axes before you describe the results addressing the question *what did you find?*
</span>
---

4. <span class="fragment " >
    <span style="color:orange">**Conclusion & Future Work**:</span> Discuss the results (your interpretation), before you list the concluding reamrks, learned lessons, and future research directions. 
</span>

---

5. <span class="fragment " >
    <span style="color:orange">**Group Discussion**:</span> This is the most important part where you need to list a few major things that you need to discuss with the group, for example: 
	- *How the paper could be improved?* e.g. critique on the proposed method, design choices, missing experiments, or inappropriate evaluation metrics. 
	- *How the paper could be applied in medical domain?* e.g. challenges in healthcare. 
	- Have a look at the reviewers feedback, if available, e.g. [openreview](https://openreview.net/)
 
</span>

---

6. <span class="fragment " >
    <span style="color:orange">**Needless to Say**:</span>  
	- Read the paper carefully, and look for complementary materials; blog posts, videos, or code repo. to better understand the paper. 
	- Be mindful of time.  You have 30 mins for points 1-4, and 15 mins for point 5. As a rule of thumb, # slides < given time slot in mins.
	- Build a compelling story and try to engage your audience. 
	- List the References in the footer of the corresponding slide 
	- <span style="color:orange">Practice, practice, practice</span>

</span>

---
## <span style="color:green">Blog Post </span> Guidelines

---
> The secret to getting into the deep learning community is high quality blogging. Read 5 different blog posts about the same subject and then try to synthesize your own view. Don’t just write something ok, either — take 3 or 4 full days on a post and try to make it as short and simple (yet complete) as possible. 
-- [Andrew Trask](https://hackernoon.com/interview-with-deep-learning-researcher-and-leader-of-openmined-andrew-trask-77cd33570a8c), DeepMind

---

This is a free-style blog post! One can hardly enforce guidelines. However, I personally liked the *Dos and Don'ts* appeared in this [blog post](https://www.fast.ai/2019/05/13/blogging-advice/).  Here some examples
- [What is Federated Learning?
](https://medium.com/@ODSC/what-is-federated-learning-99c7fc9bc4f5)
- [Federated Learning: A Guide to Collaborative Training with Decentralized Sensitive Data – Part 1](https://www.inovex.de/blog/federated-learning-collaborative-training-part-1/)
- [Do we need deep graph neural networks?](https://towardsdatascience.com/do-we-need-deep-graph-neural-networks-be62d3ec5c59)
- [FACT Diagnostic: How to Better Understand Trade-offs Involving Group Fairness](https://blog.ml.cmu.edu/2020/10/12/fact-diagnostic-how-to-better-understand-trade-offs-involving-group-fairness/)
- Other blog sites: [CMU ML Blog](https://blog.ml.cmu.edu/), [BAIR Blog](https://bair.berkeley.edu/blog/)

---
## Helpful resourses 
- [How to read a paper?](https://web.stanford.edu/class/ee384m/Handouts/HowtoReadPaper.pdf) --Three-pass method
- [How to read a research paper?](https://www.eecs.harvard.edu/~michaelm/postscripts/ReadPaper.pdf)
- [How to Read Scientific Papers Quickly & Efficiently](https://medium.com/@drewdennis/how-to-read-scientific-papers-quickly-efficiently-e7030c4018fa)
- [How to write a blog post from your journal article?](https://medium.com/advice-and-help-in-authoring-a-phd-or-non-fiction/how-to-write-a-blogpost-from-your-journal-article-6511a3837caa)
- [Advice for Better Blog Posts](https://www.fast.ai/2019/05/13/blogging-advice/)
---

![PhDComics](PhDComics.png)
Image source: [phdcomics.com](http://phdcomics.com/comics/archive.php?comicid=719). @Jorge Cham

---
Should you have any questions, please drop me an email at shadi.albarqouni@tum.de

[@ShadiAlbarqouni](https://twitter.com/ShadiAlbarqouni)