Giving a Technical Presentation
After attending and presenting at IPCV '08, I saw a number of presentations that could've been a lot better. I thought maybe the issue was that there weren't sufficient resources on how to give a technical presentation, so I did a quick google search for "how to give a technical presentation." There are 2.1 million results for that search.
I pored through some of the results and found that they are in line with how I try to approach technical presentations.
I'll break it down into a few things to do and a few things to avoid.
Things to Do
- Know your audience and tailor your talk accordingly
- Be concise
- Plan one or two goals for your audience and work to meet only those goals (in the education biz, they call these "learning outcomes")
- Know where you're presenting and what your equipment will be
- Share your passion
- Practice your talk
Things to Avoid
- Reading your slides word for word
- Turning your back on the audience
- Reading from a script
- Giving a wall of math
- Skipping around a lot
- Overusing gimmicks
The most important point in there is practice. Practice with people who you trust to be appropriately critical. Give your talk with a timer to ensure that you're on target. After the practice, write down all comments immediately (and verify that you understand the comments) and immediately address them. I'm pretty good at talking on the fly, but I'm way better if I've practiced a few times. After my first practice, I had these notes - some from me, some from my peers:
0 talk about outline - don't start talk right away
1 more about why sparse sampling - application info
2 need slide about accuracy
3 more visuals
4 include failing docs?
5 fix typo in in CoordinateGenerator
I addressed these right away. One that wasn't in my list was from Elaine - she told me directly to not read my slides. I made a mental note to work more on flow.
If you practice, you won't ensure success, but if you don't practice you will ensure failure.
If you're giving a technical talk, google says that there are 2.1 million reasons why you can give a good technical talk. Here are three: