Oral Session: Sampling from Probabilistic Submodular Models - 2016
Details
Title : Oral Session: Sampling from Probabilistic Submodular Models Author(s): Microsoft Research Link(s) : https://www.youtube.com/watch?v=kRgCUp33uLs
Rough Notes
LC-1330 Presenting Doctoral Research
Date and place/URL:
https://www.youtube.com/watch?v=kRgCUp33uLs (20 minutes)
Title:
Sampling from Probabilistic Submodular Models
Presenter:
Alkis Gotovos
First impressions (e.g. organization, overview, personal appearance, body language, target audience)
Starts with why the topic is important and why we should care - I always enjoy this especially when I am not too familiar with the topic. (Unfortunately, the video quality is not that good).
Use of visuals (e.g. balance of text and visuals, use of colour, assertion and claim evident, use of animations)
Used well, for e.g. the relation between their models and existing models was shown using a nice Venn diagram which was introduced bit by bit which is a good choice and does not overwhelm the audience.
Use of voice (e.g. pronounciation, articulation, pausing, chunking, word stress, speed)
Good, what initially looked like complex ideas were explained clearly. Any math that was introduced was also explained in clearly.
Final impressions (what did I learn from this talk - what to do or what not to do)
Great structure. What to do: Clearly delineate the work being presented and prior literature, and highlight the shortcomings of the prior work and how the presented work handles them.