It looks like this platform might finally become more populated, so let me (re)introduce this page.

Neurofrontiers is a bilingual blog about interesting neuroscience topics that tries to be accessible to the broader public while still maintaining scientific accuracy. It's run by a team of three people: a computational neuroscientist, a psychologist, and a graphic designer. We think that being on social media allows us to stay up-to-date with the most recent discussions in science and hope to be able to connect with like-minded individuals.

Posting interests below so we show up in mutual searches:

I wanted to follow someone on scholar.social but could not as I happen to be on qoto.org --- random server I picked. So I reached out to the moderators of both. Both were kind and quick to respond. What I learned might be relevant for other academics:

Scholar.social is on the more restrictive end of what they accept, and qoto.org perhaps on the other end, enough so for scholar.social to have blocked qoto. Scholar.social blocks many other servers, by their own admission, but only qoto among those catering to academics.

My conclusion, you might want to avoid both, if you don't want to be blocked from following people on either server.

The negative climate impacts of mining the cryptocurrency Bitcoin have grown rapidly over time, with carbon emissions per coin multiplying 126 times from 2016 to 2021.
During that window, the climate damage of mining one Bitcoin averaged 35% of a coin’s value, similar to the environmental costs of unsustainable products like crude oil and beef

news.mongabay.com/2022/12/clim

#ClimateEmergency #pollution #ecology #environment #ClimateCrisis #ClimateCatastrophe
#ClimateChange #Climate #Bitcoin

Learning without backpropagation is really taking off in 2022

First, @BAPearlmutter et al show in "Gradients without Backpropagation" that a single forward pass with perturbed weights is enough to compute unbiased estimate of gradients:
arxiv.org/abs/2202.08587

Then, Mengye Ren et al show in "Scaling Forward Gradient With Local Losses" that the variance of doing this is high, but can be reduced by doing activity perturbation (as in Fiete & Seung 2006), but more importantly, having many "local loss" functions:
arxiv.org/abs/2210.03310

Then Jeff Hinton takes the "local loss" to another level in "Forward-Forward Algorithm", and connects it to a ton of other ideas e.g. neuromorphic engineering, one shot learning, self supervised learning, ...: cs.toronto.edu/~hinton/FFA13.p

It looks like and are really converging.

@ct_bergstrom Maybe we need to start setting exams where we tell students to use the cheat-bots and paste in the answers, but then the actual exam is to GRADE THE BOT’S ANSWERS and explain which answers are wrong and why 🤣

We are participating in the #EEGManyLabs project - Müller et al., 2003 - Replication. We will offer this as a Master internship project to teach students at Maastricht University about #openscience and #reproducibility.

They are still looking for other labs to join! Please boost to share this more broadly!

docs.google.com/forms/d/e/1FAI

#introduction I am a professor at Penn and also co-director of the CIFAR Learning in Machines and Brains program. I like to think about neuroscience, AI, and science in general. Neuromatch. Recently, much of my thinking is about Rigor in science and I just started leading a large NIH funded initiative community for rigor (C4R) that aims at teaching scientific rigor.

My interests are broad: Causality, ANNs, Logic of Neuroscience, Neurotech, Data analysis, AI, community, science of science

Can machine learning predict risk of developing breast cancer in the future from the current mammogram? Can we use that to personalize screening?

Join us this Thursday Dec 1st at 11AM (ET) online for a conversation with Adam Yala from UC Berkeley pubmed.ncbi.nlm.nih.gov/347674

Coordinates for the event and other conversations : sites.google.com/view/ai-breas

IMPORTANT ANNOUNCEMENT!

Dear past and future NBDT authors, once your paper is accepted and has a doi (visible on the NBDT page), please go and update the arxiv version, indicating that doi and the publishing details as it appears in the api (example nbdt.scholasticahq.com/api/v1/, change the last part pointing to your paper).
In this way the bibtex entry from arxiv, and also the ones from google scholar etc, will point to NBDT and not to "misc" as any other arxiv preprint.

Thanks!

Here is a machine learning challenge: "Decode" EEG to estimate what a listener was hearing, actually, what features of continuous speech can be predicted from EEG.

exporl.github.io/auditory-eeg-

The challenge runs from now until February 6, 2023, after which the top 5 teams will be invited to submit a 2-page paper to ICASSP and later on an invitation to write a journal paper for the IEEE open journal of signal processing.

3,013 neurons, half a million synapses: the complete #connectome of the whole #Drosophila larval brain!

Winding, Pedigo et al. 2022. "The connectome of an insect brain" biorxiv.org/content/10.1101/20

We’ve mapped and analysed its circuit architecture, from sensory neurons to brain output neurons, as reconstructed from volume electron microscopy, and here is what we found. 1/

#neuroscience #connectomics #vEM #volumeEM

RT @sheacshl This newly posted work from @RJordan and Georg Keller looks super cool! Locus coeruleus responds with a widely broadcast visual feedback prediction error that acts as a teaching signal for sensorimotor learning. biorxiv.org/content/biorxiv/ea

What happens when you use #autodiff and let your nonsmooth iterative algorithm goes to convergence?

With J. Bolte & E. Pauwels, we show that under a contraction assumption, the derivatives of the algorithm converge linearly!
Preprint: arxiv.org/abs/2206.00457

I will present this work this week at #NEURIPS2022

After two years of negotiations with Microsoft, the joint committee of the German federal data protection authority and 17 state regulators (DSK) published a devastating statement that essentially says that organizations currently cannot use MS365 in a lawful way under the GDPR.

Show thread

Unlike @PessoaBrain, we argue that reductionism remains a valuable and viable approach to understanding brain function.
Like him, however, we suspect there is a better way (not emergence but expansion; preprint link to "Disentangling the brain w/ Srirangarajan)...
stanford.edu/group/spanlab/Pub

Hello Mastodon peeps 👋! This is what it looks like across hundreds of neurons in prefrontal cortex when a monkey starts dozing off.

I'm working with an organization that may eventually fund proposals to fund workshops for research groups working on "mathematics for humanity". This would include math related to climate change, democracy, economics, health, maybe AI risks, etc.

I can't give details until it solidifies.

However, it would help me to know a bunch of possible good proposals. Can you help me imagine some?

A good proposal needs these things:

(1/n)

Show older
Qoto Mastodon

QOTO: Question Others to Teach Ourselves
An inclusive, Academic Freedom, instance
All cultures welcome.
Hate speech and harassment strictly forbidden.