I am available. Happy to present
1 Like
Thanks @janardhanp22 . I think it would be great if @samar and @singlasahil14 could present remotely! 
Works for me too! Samar also mentioned that it works for him. Need hear back from Sahil yet.
@samar that’s great! Make sure you have a good quality headset and skype connection in a quiet location for the presentation please
Yes I am in. Will create a good presentation next time
Hi @jeremy, @sravya8. I can make it, have we given any thought to how we can improve on the last session. I thought some talks felt a little rushed and deserved more time (like the urdu language talk). Also, should we have the remote speakers pre-record their talks to a given time limit for better control over quality? I felt bad there were so many technical distractions from the actual content.
2 Likes
@jeremy Can we have a second session as well? Two reasons to ask:
- We have three more speakers who might be interested: @brookisme, @Dario, @xinxin.li.seattle
- Splitting into two sessions would give us an opportunity to have more time than 5 minutes per speaker. I personally think it would be better to have at least 7-8 minutes for each talk as it is important to cover some domain knowledge as well as technical approach.
@sara.hooker13, @samar, @singlasahil14, @janardhanp22, @ljubomir what do you guys think about time per talk?
I am with you for having more time per talk, especially given that each talk needs to cover some domain knowledge and impact potential along with technical details to be useful.
With respect to pre-recording remote talks, it is really up to what remote speakers think. @samar, @singlasahil14 ?
@sravya8 I support allowing more time for each talk, if logistics allows. Five minutes is just too short for a useful presentation given that in this audience not everyone may be familiar with deep learning
@sravya8 I suspect it will depend on how the first one goes! I’m sure if it’s a success (which is extremely likely…) the Data Institute would want to do a follow-up. I’m happy for you to curate this first one however you think is best, in terms of content and timings - this is your baby! Perhaps you can suggest which talks we should start with, how many, and how long? I’ll ask David and Mindi about whether there are other dates free.
@sravya8 I’m okay with it either way, but I think playing a recorded video might feel a bit disconnected and not quite the same as having a speaker present in real-time. Maybe we can do a test run right before to ensure everything is working smoothly. Do let us know what you think sounds better!
@sravya8 I’m all for increasing the duration. ~8 minutes sounds good, since we first have to introduce the audience to the background of the problem and then also discuss how the models we use to solve it actually work.
1 Like
I agree live talks are more engaging. Test run sounds fine. In terms of technical logistics, I guess its better if we keep it simple this time as it is mainly geared towards in-person attendees. We can use hangouts just for remote speakers? Everyone else just connects their laptops to the projector. Bonus: Maybe one of us in the audience can just do a facebook live from our phones?
@jeremy FYI: Just heard back from Prof. Yannet Interian and she says 5/26 is available and she will block it for us. If that gets finalized, I think 5 talks ~12 minutes each should be good for the first session. ~8 minutes talk and ~2-4 minutes Q&A or something like that. Others: let me know your thoughts.
@sravya8 Yes it will be difficult to drop the speakers instead if we can get another slot then we can accommodate everyone. Since it is a seminar from USF, I am confident that there will be too many technical difficulties as last time and we can move from one speaker to another in a faster manner.