Source of Give a talk at DLCT


  [Deep Learning: Classics and Trends (DLCT)](https://mlcollective.org/dlct/) is our longest-running research event, having been occurring weekly since 2018. If you have been invited, and scheduled to give a talk there, here is what to expect. (If you would like to but have not yet received an invitation, [nominate yourself here](https://docs.google.com/forms/d/e/1FAIpQLSdoohVxHfu0DNPlCHJ-GpPnQz9C9E9GmMeA3sL0nigFFrjlQQ/viewform?usp=sf_link)!)

A few casual definitions of what DLCT is:

 - a journal club (where someone presents a paper)
 - a virtual group meeting (on Zoom, where anyone can unmute and talk) that's radically open (anyone can join)
 - an open-mic stage to try out new talk ideas (where the sample of audience is large, unbiased, and best of all, friendly)
 - a safe place (we don't record)
   
## Time, length and format 
The event is from 10 to 11am, Pacific Time, every Friday. To help with timezones, view it on MLC's [events calendar](https://calendar.google.com/calendar/embed?src=e4p5s7715ersqsoet6cq4509q0%40group.calendar.google.com). 
 
The presenter will have 50 mins (since the host will take about 10 mins in the beginning to open up) that they can use however they'd like, but we recommend preparing for no more than 35 mins of content, as one distinct trait of DLCT is that it's _highly interactive_. Think of it as a hybrid between a presentation and a conversation. Other characteristics of DLCT are “_casual_," "_free-flow_" and “_unfettered_.” **We do not record**, to allow people to “_ask stupid questions_,” which we found to be the best way to connect and learn. It is common for people to stick around for up to 30mins after the hour mark to engage further with the speaker, but it is not required for the speaker to commit to staying overtime.
 
Occasionally, though, we are open to recording the talk in the case of high demand and specific requests. This [YouTube watch list](https://youtube.com/playlist?list=PLfeYlJzwvDN3HmOGKtB78VjkB0W7E7Sy8) contains those occasional recordings. If you are the presenter and would like your talk recorded, kindly let the host know beforehand.

The meeting allows up to 100 attendees, first come first serve. Attendees are able to unmute themselves and speak up; there’s also a chat window where they might post questions. It will be lightly moderated by the host.

## Audience
In terms of audience, we announce the talk to a [mailing list](https://groups.google.com/g/deep-learning-classics-trends) of ~4000. In the past sessions we have had on average 80 attendees each time. Since it is an open signup, people from all kinds of backgrounds attend. The questions and discussion so far have been of quite high quality.

## Scope of the talk
We do want you to go deep on technical details! Those whom we invited are usually aligned with our research interest, so if I were you I'd just go as deep as needed, technical wise, and assume that people have enough ML/DL background. That said, a couple of slides in the beginning that zoom in from the grander scheme of ML to the specific problem(s) you are trying to solve would help engage the audience.
 
## Short interview
The host would like to do a short (5-8 mins) interview with you, prior to your talk, just to warm it up, and give the audience a chance to know more about you. Please let her know 1-2 questions from below that you'd like to be asked. Also feel free to propose questions yourself! If you want to opt out of this interview it is totally fine too.
 
Example questions:

1. What's one thing people would be surprised to know about your research, or about you?
2. What are bad recommendations or advice you hear given in your field?
3. What's a life hack, or an unusual habit, you have or are hoping to develop?
4. What is an overlooked element that makes a piece of research stand out?
5. What are unknown struggles in becoming a researcher?
6. What is an underrated skill for an ML researcher?
7. What is the one advice you would give your past self when you were just starting in research?

## What we need from you before the talk
The host, Rosanne, announces the talk of the week one day ahead — on Thursday morning, with a **title** and **abstract** of the talk, along with a **bio** of the speaker, and **links to the paper(s)**. If you can send them to her by Wednesday night, that'll be great. No worries if your talk is still far away; you will get a reminder when it is closer. If you opt in for the interview, please send Rosanne your selected **1-2 questions** before the talk starts!

## What we need from you after the talk
Slides, if you are willing to share, will be posted on the page post your talk. Please send slides (PDF or a link that you host yourself) to Rosanne when you have a chance. If it had been a recorded session, let us know your editing requests too!

## Promotion 
We sometimes promote the talk info on the day or a few days ahead, using either the host's twitter handle ([@savvyRL](https://twitter.com/savvyRL)) or that of MLC ([@ml_collective](https://twitter.com/ml_collective)). Feel free to tag either/both if you are promoting the event yourself. Please do not share the zoom info to the public, and instead point the audience to the [event's website](https://mlcollective.org/dlct/), where they are asked to sign up the [email group](https://groups.google.com/g/deep-learning-classics-trends) and find the zoom info from the most recent message. They can also navigate to the [MLC Events Calendar](https://mlcollective.org/events/#cal) where Zoom info is displayed inside of the corresponding event description.

## Recognition
As ML Collective is a **registered 501(c)3 non-profit** in the United States, giving a talk at DLCT is a charitable contribution to society. If your employer has a system for tracking, logging, and recognizing such volunteering activities, please go ahead and log however many hours you spent on this talk (from preparation all the way to delivery). It will also help with our reputation by having you on record as a valuable volunteer! If there’s anything else we can do to help with the recognition of your time spent, please let us know.

## Misc
More information and all past and upcoming talks can be found [here](https://mlcollective.org/dlct/). You are very welcome to sign up the [listserv](https://groups.google.com/g/deep-learning-classics-trends) to tune in for any session, any Friday. It might help you get a sense about the vibe of the audience there.
 
That's it. Thanks for reading to the end! :)