SVRHM 2022 Workshop
@svrhm_workshop
4th @NeurIPS2022 Workshop: Shared Visual Representations in Human and Machine Intelligence. Organizers: @ArtDeza @joshuacpeterson @apurvaratan @cocosci_lab
ID: 1154773354018217984
https://www.svrhm.com 26-07-2019 15:19:57
384 Tweet
670 Followers
886 Following
A relevant Mid-Level Vision Toolbox developed by Dirk Bernhardt-Walther is on Bluesky 's group that popped up during his talk earlier today! : github.com/bwlabToronto/M…
Come see our work on Homomorphic Self Supervised Learning today at the NeurIPS Conference SSL workshop! Stemming from my internship at Apple with Xavier Suau and Luca Zappella, we explored the synergistic relationship between equivariant backbones and SSL: arxiv.org/abs/2211.08282
🧵 Introducing Torchlens, a package for extracting & mapping the results of every intermediate step in any PyTorch model in 1 line of code ("every" & "1" mean just that; "any model" is the goal 🤞) 1/ GitHub: github.com/johnmarktaylor… CoLab Demo: colab.research.google.com/drive/1ORJLGZP…
It was great to meet in person again at @SVRHM_workshop! In this thread, I briefly describe our SVRHM paper (openreview.net/forum?id=a3YPu…), for which Wenxuan Guo (equal-co-1st), Heiko Schütt, Nikolaus Kriegeskorte, and I won the Common Sense Machines best paper award. 1/13
Looking forward to reading this paper from Simon Kornblith and colleagues on human alignment of neural network representations: arxiv.org/abs/2211.01201 tl;dr: objective fn and dataset have a larger impact than architecture and scale, so scale isn't all you...
