Carlos Stein N Brito

Computational neuroscience, ML theorist. Um olho no peixe, outro no gato

[Carlos Stein N Brito @cstein06]

@royalserf pure Adam Curtis

[Carlos Stein N Brito @cstein06]

See you all soon at Cosyne!!
https://t.co/pkDfQHEC0s #cosyne2025 #cosyne25

[Carlos Stein N Brito @cstein06]

@Grimezsz Oxytocin might modulate some empathy drives up and technical down

[Gerstner Lab @compneuro_epfl]

Most methods of sparse coding or ICA assume the 'pre-whitening' of inputs. @cstein06 shows that this is not necessary with a smart local Hebbian learning rule and ReLU neurons!
Paper just out in @PLOSCompBiol:
https://t.co/d3cFN4oxOm

[CosyneMeeting @CosyneMeeting]

Last spotlight of Cosyne Day 2 🔦

Meet @cstein06 our Audio-Video Chair (thank him for all the video recordings 🙏)

This is his 11th 🤯 Cosyne in a row, and it holds a special place in his heart bc it is around his birthday!! His first Cosyne talk was in Salt Lake 10 years ago!

[CosyneMeeting @CosyneMeeting]

Chairs lunch!!

If you are enjoying your time at #Cosyne24 you should thank these folks when you see them around!

From the program, to the workshops, to the video recordings, to the grants, to so so so much more we don’t see 👏👏👏

[CosyneMeeting @CosyneMeeting]

Reminder that the main meeting is live streamed!!
#Cosyne24

https://t.co/9bJw7r1Qtf

[Carlos Stein N Brito @cstein06]

@TonyZador Initial turning points: you can use a=b as the initial point, to get c as the 2nd turning point. But I don't see how to get first turning point, would need to numerically sweep. Starting from 0. gets: (0., 2.5, -26.). From -1: (-1, 3.1, -45).

[Carlos Stein N Brito @cstein06]

@TonyZador Given it's zig zag, for a (a,b,c) sequence of turning points, you can optimize the cost for b, take derivative, equal to zero, and get something like: pdf(b)*abs(b-c) - (1 - abs(cdf(b) - cdf(a))). From this, given a,b first turning points, you optimize c, and get a sequence.

[paulo moreira 🚴‍♂️ @paulomoreria]

diga pode dizer o que foi

[Carlos Stein N Brito @cstein06]

@emollick 7 Billion Humans!

[Carlos Stein N Brito @cstein06]

@neuro_kim And follow the live video streaming on the Cosyne Youtube channel! https://t.co/S1Y4mdJ5pc

[Kim Stachenfeld (neurokim.bsky.social) @neuro_kim]

Tremendously excited to be giving an RL tutorial at #cosyne2023!

Materials for the tutorial can be found here.

slides
https://t.co/2Iv4njlGNr

coding colab (File > Make a copy; Runtime > Change runtime type > GPU)
https://t.co/QZjb4NPq1w

[Carlos Stein N Brito @cstein06]

Cosyne 2023 Tutorial with Kim Stachenfeld has started! https://t.co/VbkeX1YC0D

[Carlos Stein N Brito @cstein06]

@rs_williamson

[Carlos Stein N Brito @cstein06]

@jmourabarbosa @CosyneMeeting What? Let's go?!

[Carlos Stein N Brito @cstein06]

@GothamChess What's their performance Elo rating?

[Carlos Stein N Brito @cstein06]

@Samuel_Eckmann @DrYohanJohn @PessoaBrain @NoahGuzman14 Thanks for the mention. Might also be relevant here, in this recent paper we explain the functional form of Hebb rules (including BCM, triplet-STDP) as learning higher-order components (i.e. ICA, sparse coding) while ignoring second-order information: https://t.co/u7pPYy2yIs

[Julijana Gjorgjieva @GjorJulijana]

Don't forget to apply to the Cajal course in Computational Neuroscience! Deadline March 28!👇