Scheduling meeting for next week

[Stephen Roller]

Hey guys,

Let's have a FLARE meeting next week. Fill out this doodle. Note that I'm ruling out Monday's altogether because I have enough Monday stuff already.

http://doodle.com/poll/27wvkzb3gwb64d2r

I think everyone agreed that we want to use FLARE this semester to review some of the big trends from the past 6-12 months in deep learning, focusing only on stuff that's since been BIG.

Here's a few proposals, and does anyone want to volunteer to go first? Your responsibility would be to do the lit review to identify the 1 or 2 seminal papers for the topic.

- Ultra Fast Deep Learning Intro (Unclear if we want this?)

- Highway networks/Shortcut connecdtions

- Adversarial Networks

- LSTMS w/ Attention

- Memory Networks

- Deep Q learning

- Bayesian Deep Learning

- What else? What's the major development in ConvNets from the past year?

Cheers,

Stephen

[Stephen Roller]

I realized a bunch of people hadn't gotten this email because they weren't on the list yet! Repeating this message for their benefit.

[Wesley Tansey]

I'd be willing to lead a discussion or two on Bayesian deep learning stuff at some point. The two main things going on here are: 1) the generative interpretations of various NN architectures like ReLUs (which turn out to correspond to truncated Gaussians), along with the associated scalable MCMC algorithms like stochastic MCMC; and 2) cutting-edge things like all this "normalizing flow" stuff that's coming out where you try to make variational inference algorithms that can actually approximate very complex posterior distributions to the point of being a first-class citizen almost akin to MCMC. The former is mostly coming from Lawrence Carin's lab at Duke; the latter is generally coming from Deep Mind. Happy to present one or both, but probably not all in the same week.

--
You received this message because you are subscribed to the Google Groups "FLARE" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ut-flare+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Joel Iventosch]

I am really interested in LSTMs w Attention. I probably don't have enough experience w them to lead a discussion on them at this point (at least to the level that some others might be able to), but definitely a vote of interest for that topic, and I'd be happy to do some reading/lit survey to help tee that topic up.

[Aishwarya Padmakumar]

I guess I second Joel. LSTMs with attention is probably familiar to bunch of people here but it's something I really feel I need to know more about and haven't really looked at much before.

Also, I'm still not getting mail from this group although I've been added to it. Anyone has any idea why that could be happening?

On Monday, August 29, 2016 at 12:10:56 PM UTC-5, Stephen Roller wrote:

[Nazneen Rajani]

Change your notification settings for this Google group, I was facing the same issue..

--
You received this message because you are subscribed to the Google Groups "FLARE" group.

To unsubscribe from this group and stop receiving emails from it, send an email to ut-flare+u...@googlegroups.com.

[Stephen Roller]

Hey everyone, please fill out the doodle by the end of today! I’ll make an announcement later.

Cheers,

Stephen

---------- Forwarded message ---------
From: Stephen Roller <m...@stephenroller.com>
Date: Mon, Aug 29, 2016 at 12:10 PM
Subject: Scheduling meeting for next week
To: ut-flare <ut-f...@googlegroups.com>

[Md. Mustafizur Rahman]

Can someone please announce the location of the meeting?