4 ms·

My recommendation is: start with an image (no text) on the first slide. It is important that the image has seemingly nothing to do with the topic of the talk th

by jll29 2y ago

My recommendation is: start with an image (no text) on the first slide. It is important that the image has seemingly nothing to do with the topic of the talk that was on the (unnumbered) title slide. Now people are curious what explanation you are giving, and are paying attention. After resolving the mystery, you can move on to the second slide, you can give e.g. a problem statement or research question and then follow with the usual structure (outline, method, data, experiments, evaluation results, discussion and limitations, summary, conclusions and future work. Caveat: This works only for oral presentations; there is an important type of slide deck with is the predominant type in large international corporations, and it's a mix between PPT presentation and Word document, where slides are full of text so that the deck is self-explanatory. Those are written not just to be presented but MOSTLY for being passed around by email to be read. They intentionally violate all the rules of good slide decks to accompany good presentations, because they are mainly used to summarize e.g. a new business case for senior management, who may never listen to the presentation, but who may browse the slides.

dimatura 2y ago
There's a related medium where I think similar advice applies, technical papers. At least for my field - computer vision / machine learning. You choose a cool, hopefully self-explanatory, figure to feature prominently in the first page. Something that will grab the attention of someone flipping through pdfs and get them hooked. In computer vision you can usually find something visually appealing, like a 3D reconstruction, or an image with highlighted object detections. If nothing else, you can go with a plot showing how much better your method is compared to a baseline, but that's probably less engaging for people who aren't super familiar with how meaningful the numbers are.
interludead 2y ago
The hardest thing sometimes is to capture audience’s attention and spark curiosity

by jll29 2y ago

dimatura 2y ago

There's a related medium where I think similar advice applies, technical papers. At least for my field - computer vision / machine learning. You choose a cool, hopefully self-explanatory, figure to feature prominently in the first page. Something that will grab the attention of someone flipping through pdfs and get them hooked. In computer vision you can usually find something visually appealing, like a 3D reconstruction, or an image with highlighted object detections. If nothing else, you can go with a plot showing how much better your method is compared to a baseline, but that's probably less engaging for people who aren't super familiar with how meaningful the numbers are.

interludead 2y ago

The hardest thing sometimes is to capture audience’s attention and spark curiosity