WIP: What happens to intermediate representations as more layers are left at their random initialization? Here: the similarity between a convnet trained normally (all layers trained) and networks where only layers after layer n are trained (early layers untrained).

After this last season of #GameofThrones

Der WAHL-O-MAT der bpb.de ist ausgefallen, die PARTEI bietet Ihnen den PARTEI-O-MAT: gleicher Inhalt, cooleres Design, leichte Benachteiligung von unseriösen Kleinstparteien... Hopp, auf! ZwinkerSmiley partei-o-mat.de

I have zero problem with people reporting results they found with p=0.1. Or p=0.2. Whatever. The problem is with statmodeling.stat.columbia.edu/2019/05/26/i-h…

PACMAN in the news again: Depth-resolved ultra-high field fMRI reveals feedback contributions to surface motion perception biorxiv.org/content/10.110…

This might be the perfect overhyped #AI meme. Courtesy of Chris Russell

Question from the audience: "Could Karl Friston minimise my free energy by speaking a bit louder please?" I couldn't agree more #ccn2019 #ccneuro2019

"Due to random chance, some percentage of models will outperform other ones, even if they are all just as good as each other. Maths doesn’t care if it was one team that tested 100 models, or 100 teams." lukeoakdenrayner.wordpress.com/2019/09/19/ai-…

Solid defense of the insightful contributions for the better understanding of the conscious motion perception in humans at ultra-high field fMRI by Dr. Marian Tough questions by the corona. Congrats! #PhDChat #hMT #UHF #neuroscience #consciousness #sciencetwitter #ColumnarfMRI

Scientific activities sorted into Harry Potter houses: Gryffindor (courage, boldness) giving talks Hufflepuff (hardworking, honest) writing and reviewing papers Ravenclaw (intelligence, cleverness) programming and statistics Slytherin (ambitious, cunning) grantwriting

If I was ever gonna be part of a production. This would be the only way.

In preparation for #OpenMRB2020 , happy to announce my *novel* algorithm: skullripper2020 . For those who want to strip skulls 💀 with style ⚡️ ...

New layer-fMRI paper on feedback contributions to visual surface perception. doi.org/10.7554/eLife.…

We've announced the full cohort from Interfaces Camp 👉👈 Read about all 10 companies here: betaworks.com/writing/ai-cam…

Just a few highlights from Demo Day last week with nubrain Presq Adam Mark Saleh #bunnysniffs

Other than Spotify Wrapped, what’s your favorite way to use AI? If you’re struggling for an answer that's not "chat", today’s your lucky day – we've got footage from Interfaces Camp IRL-only Demo Day last month (not included two teams in stealth mode) Thread w. videos 👇👇👇

For building “mind reading” technology, the huge differences between EEG devices / datasets are a major challenge. This paper might have the solution: brain-bzh.github.io/reve/ 🧠

🚀 nubrain raised $565K pre-seed led by Betaworks with support from @TrueVentures, RRE Ventures Malte Ubl (CTO, Vercel), Dimitris Tsementzis (Head of Applied AI, Goldman Sachs), and Philippe Brule (CTO, Restack) Goal: Make thought a native input to AI. 🧠✨