Incredible to have so many amazing Lousianan memories in the past 1.5 year: TubeDETR, FrozenBiLM, VidChapters, booth slots, chance encounters, swamps, plantations, NBA, po boys, jazz bars and so on… #NeurIPS2023 #NeurIPS2022 #CVPR2022

keyboard_arrow_left Previous keyboard_arrow_right Next

[ML&Physics challenge (1/..)] Call to participation: Machine Learning for physical simulation Based on the AirfRANS dataset proposed by Extrality and @Sorbonne_Univ_, published at #NeurIPS2022 …l-simulation-challenge.irt-systemx.fr Dataset: openreview.net/forum?id=Zp8Ym…

@alexschwing @ECEILLINOIS @IllinoisCS Impressive #NeurIPS2022 contributions! Apprehensive to explore further on PAC-Bayesian Bounds, Animal Reconstruction, Invariant & Equivariant Deep Nets, GAN training with limited data, and RL from demonstrations. Eager to learn about specific methodologies employed.

We scale to deep neural networks. Similar to our last year's #neurips2022 LILA method for invariance learning, we utilise modern linearised Laplace approximations (KFAC) to estimate differentiable marginal likelihoods, which encourage wide minima.🧵 6/11 x.com/tychovdo/statu…

Tycho van der Ouderaa @tychovdo

2 years ago

We scale to deep neural networks. Similar to our last year's #neurips2022 LILA method for invariance learning, we utilise modern linearised Laplace approximations (KFAC) to estimate differentiable marginal likelihoods, which encourage wide minima.🧵 6/11 x.com/tychovdo/statu…

4 48 261 0 135

Download Gif

I did a quick analysis of #NeurIPS2022 authors, and it looks like Google is the winner of this conference - out of 5261 unique authors, 220 of them belong to Google+DeepMind (so over 4% of all of them!) Welcome to a short thread about arguably the most well-known AI conference

At #NeurIPS2022 we presented Variational Interpretable Concept Embeddings (VICE), an approximate Bayesian inference model for inferring mental representations of object concepts from human similarity judgments. proceedings.neurips.cc/paper_files/pa…

i've just finished my role as a program chair of #ICML2023 after serving in the same role earlier for #ICLR2020 and #NeurIPS2022. i had so many thoughts on what i would do when this actually happened, but now that it happened, i can also say ..

How to improve your self-supervised learning performance and unify the various SSL theories? Predict your augmentations! That's the gist of our #NeurIPS2022 paper S-TEC (see more in 🧵below 👇). We also showed how Efference Copies in the biological motor system could play a…

The majority prefers the "no orals" single-track format of #CVPR2023😯 #NeurIPS2022 also had no orals. So is this the future? are orals 'dead'?

#CVPR2024 @CVPR

10 months ago

The majority prefers the "no orals" single-track format of #CVPR2023😯 #NeurIPS2022 also had no orals. So is this the future? are orals 'dead'?

2 5 41 13K 2

Download Image

@tsay_jonathan @ayirpelle @akaijsa @pranavm42 @eLife It's a really good question; we focused on neural predictions but we don't make explicit behavioral predictions. But some work from @TrackingPlumes is going down that path/ his #neurips2022 paper (although this is more algorithmic investigation) arxiv.org/abs/2209.14218

#ICLR2023 참석 중인 한국 학생분들, 오늘 4시 45분에 학회 호텔 (radisson) 1층 바에 오셔서 저랑 맥주 한 잔 하시죠. 특히 #neurips2022 랑 #icml2023 리뷰 하신 분들은 약속 한대로 맥주 한 잔씩 제가 삽니다.