ΑΙhub.org
#NeurIPS2021 in tweets – highlights from the first two days
The 35th conference on Neural Information Processing Systems (NeurIPS2021) started on Monday. So far, attendees have been enjoying the tutorials, the affinity group workshops, and the new datasets and benchmarks track. Participants are looking forward to the invited talks, poster sessions, and workshops that are coming up during the week. This compilation of tweets gives a flavour of the conference so far, and events to look out for later in the week.
Tutorials
Welcome to Day 1 of #NeurIPS2021 🎉 Today is all about tutorials. Our blog summarises highlights for day 1. https://t.co/kNKjujY5r4
— NeurIPS Conference (@NeurIPSConf) December 6, 2021
If you’re at @NeurIPSConf @cephaloponderer and I have this tutorial tomorrow (Monday). We will also be joined by @alexhanna and @willie_agnew for the extended Q&A/panel—we are so thankful that they are making time for this in their extremely busy schedules. https://t.co/eMG8WdwCcB
— @timnitGebru@dair-community.social on Mastodon (@timnitGebru) December 5, 2021
We’re excited to be ‘here’ at #NeurIPS2021 and making the most of today’s tutorials, which are free for all🤩
Pictured is @vukosi and @davlanade's “A Journey Through the Opportunity of Low Resourced Natural Language Processing — An African Lens” pic.twitter.com/CU97nQwosi
— AIhub (@aihuborg) December 6, 2021
Awards
Paper awards for @NeurIPSConf have been announced!🎉#NeurIPS2021 https://t.co/AEiFTXhKFQ
Congrats to all the winners, I'll link to the Outstanding Paper Awards 🧵
1. A Universal Law of Robustness via Isoperimetry, by @SebastienBubeck & @geoishard.https://t.co/npXQbfkPbP (1/n) pic.twitter.com/qYHLcyIyGX
— Gautam Kamath (@thegautamkamath) November 30, 2021
Workshops
Save the date!📢📅 Our workshop, Tackling #ClimateChange with #ML, is happening at #NeurIPS2021 on Dec 14!
To follow all of the workshop events check out the full program at https://t.co/QeiLvu2RIW pic.twitter.com/fmRfe6cEBC
— Climate Change AI (@ClimateChangeAI) December 6, 2021
StyleCLIPDraw: Coupling Content and Style in Text-to-Drawing Synthesis
Their results suggest that the style of the drawing is coupled with the drawing process itself.
This work will be presented as an oral at our NeurIPS creativity and design workshop!https://t.co/gmf54vh3wy pic.twitter.com/o6HpDgnErp
— hardmaru (@hardmaru) November 8, 2021
Affinity groups
✊🏾 The BAI Workshop is coming! Meet one of our keynote speakers: Louvere Walker (@LWH_Bos), MathWorks Application Engineering Team Leader. Louvere will speak on the topic: What does “AI for social good” mean for the African diaspora?
More info: https://t.co/wxd4GuUl4d pic.twitter.com/urpgk0uYxi— Black in AI (@black_in_ai) November 30, 2021
Three more days until the @_LXAI at #NeurIPS2021! get ready to join the Mentorship Hour on December 7, 2021 at 4:30 PM PST. Register now: https://t.co/TlBfRAIB0P@NeurIPSConf #lxai #ai #NeurIPS2021 pic.twitter.com/PimGJXUSWR
— LatinX in AI (LXAI) @ CVPR 2024 (@_LXAI) December 4, 2021
Very honoured by the invitation. Looking forward to the @WiMLworkshop at #Neurips2021 🥰🌲🥰🌲🥰🌲 https://t.co/SdWPO3lmeY
— Meire Fortunato (@FortunatoMeire) December 5, 2021
Really excited to have been invited to be part of the panel on Towards Animal Centric AI tomorrow to talk about how we can help mitigate risks for animals when using AI! https://t.co/x7WbGAEr4J #NeurIPS2021 #QueerinAi
— Luisa Ruge PhD (@AnimalCentered) December 6, 2021
Posters
looking forward to #NeurIPS2021 starting tomorrow!
not sure what the point of posters are with virtual conferences, but i decided to have fun with two of mine:
L: https://t.co/KkFLcayyiZ
R: https://t.co/WkFMPrSy8fsee you during @NeurIPSConf ! pic.twitter.com/jt0xctPdn2
— Pablo Samuel Castro (@pcastr) December 5, 2021
Papers
Q. What does Noether’s theorem tell us about the “geometry of deep learning dynamics”?
A. We derive Noether’s Learning Dynamics and show:
”SGD+momentum+BatchNorm+weight decay” = “RMSProp" due to symmetry breaking!w/ @KuninDaniel#NeurIPS2021 Paper: https://t.co/vNU3JgHNhv
1/ pic.twitter.com/PflV4MXIyG— Hidenori Tanaka (@Hidenori8Tanaka) December 6, 2021
So much of ML focuses on "zoo conditions", with iid and trustworthy data.
Our #NeurIPS2021 takes learning 𝕚𝕟 𝕥𝕙𝕖 𝕛𝕦𝕟𝕘𝕝𝕖. We construct optimal algorithms, and prove impossibility theorems on what any algorithm can achieve in real environments.https://t.co/ntTcKB52Qg pic.twitter.com/rutr1gKjD4
— Lê Nguyên Hoang (@le_science4all) December 6, 2021
Getting ready to present
In the office early setting up for our #NeurIPS2021 oral on Learning to Draw pic.twitter.com/U4yM7evFsj
— Professor Jonathon Hare (@jon_hare) December 7, 2021
Datasets and benchmark track
Just skimmed the #NeurIPS2021 Datasets and Benchmark papers abstracts and I want to highlight these papers introducing natural language datasets which name the languages up front (title or abstract):
— @emilymbender@dair-community.social on Mastodon (@emilymbender) December 5, 2021
Panel discussions
Consequences of Massive Scaling in Machine Learning opening panel #NeurIPS2021 agrees LLMs haven't yet shifted paradigm, but might do so when we find emergent intelligence *without* understanding. /w @MelMitchell1, Noah Goodman, @OriolVinyalsML, Joelle Pineau, Jared Kaplan pic.twitter.com/q9yPFIoPy6
— Zeta Alpha (@ZetaVector) December 7, 2021
Also, holy shit that panel was awesome. #NeurIPS2021 https://t.co/258MyN2MrT
— Jade Abbott (@alienelf) December 7, 2021
A mirror conference in Switzerland
Thank you @KrzakalaF @zdeborova and Nicolas Flammarion for organising this true #NeurIPS2021 mirror experience at the beautiful Swiss convention center of @EPFL pic.twitter.com/2mqecBdhuL
— Claire Vernade (@vernadec) December 6, 2021
The virtual environment
#NeurIPS2021 avatar creation is hilarious! Conferences are actually just gameboy now. pic.twitter.com/DhhbkcQgET
— Arsha Nagrani (@NagraniArsha) December 7, 2021
How It Started #NeurIPS2017 vs. How It's Going #NeurIPS2021 @NeurIPSConf 😂😂 https://t.co/PCtvOZC8g5 pic.twitter.com/uh8M4kGj17
— Bhavya Kailkhura (@bkailkhu) December 6, 2021
tags: NeurIPS, NeurIPS2021
AUAI is supported by:
