A new preprint! Bartlett and Welch pioneered stats for estimating frequency spectra in the 50s and 60s. Their core methods are still widely used today though statistical methods have greatly advanced.
We blend Welch-style spectrum estimation and general linear modelling in a single framework allowing contrast coding, confound regression, multi-level modelling and non-parametric permutation.
The GLM-Spectrum:
https://www.biorxiv.org/content/10.1101/2022.11.14.516449v1
@ajquinn this is cool, love the idea, intuitive and useful!
What math do neuroscientists need to know?
A highlight of #SFN2022 was Ella Batty's answer to this question. She showed off an incredible math for neuroscientists course she has developed at Harvard with open materials (https://ebatty.github.io/MathToolsforNeuroscience/) and discussed her amazing work with Neuromatch Academy (https://compneuro.neuromatch.io/)
Her SFN slides are here: https://osf.io/s94b2
Spikes, Local Field Potentials, and Waves with Leslie Kay, Earl Miller, and Tony Zador
https://youtu.be/qftGEaWKTfI
Because this was my only post ever that got +5k likes on Twitter, it is only fitting that this is my first post here ⤵️
I created an awesome-PhD list on GitHub where everybody can contribute with their own tools and resources! 🔥
✨ Check it out and contribute yourself via pull requests: https://github.com/helenahartmann/awesome-PhD ✨
The #fediverse is one of the most hopeful aspects of the internet, since it tries to bring it back to its decentralized, federated roots. In my opinion, there is no other way to guarantee free communication.
However, it is not without its challenges. Looking ahead, it's easy to identify some of them:
1. Centralization is an emergent phenomena. We can see this happening with the distribution of instance sizes, and the clear “rich get richer” dynamics that take place. This is not due to bad actors, it is simply how people tend to behave.
2. Commercialization is yet to hit this platform, but if it becomes popular, it will undoubtedly happen. Instances that serve ads, promotes content, etc, will appear at some point.
3. Related to point 2, it is possible that Google and/or Microsoft will seise the opportunity to try once more to move into the social media domain, and make services that connects to the #fediverse. This will be sold as an adoption of the federation, and hence progress. But to see how that goes, we need only to remember what happened with google chat and the open Jabber protocol — they abandoned it as soon as their service became popular enough. Bait and switch.
4. Legal intimidation of smaller instances is a real menace. The push to force Facebook/Twitter/etc to be liable for what users post will backfire, and render every instance admin responsible for what every single user posts. Note that it does not matter one bit if the content is actually problematic — a single cease or desist letter from a lawyer will terrify most admins out there.
These are not problems without solution, but it's unclear how things will unfold.
.@amykooz presents some interesting work on how the brain energy landscape changes in the LSD state: a flatter energy landscape leads to increased entropy in the brain #brainmodes2022 https://www.nature.com/articles/s41467-022-33578-1
#introduction I am a postdoc At Boston University/Mass General Hospital interested in the statistical modeling and analysis of neural rhythms and their quirks, particularly real-time estimation and networks. Currently deep in studying sleep rhythms and how their disruption in epilepsy might explain some of the co-morbid cognitive dysfunction while also figuring out how to accurately assess phase for the funky waves the brain presents us with.
Also: looking for a job for next year in Europe!
PostDoc At Boston University. Sleep | Rhythms | Networks | Photography | K'taka ಪ್ರೇಮಿ www.anirudhwodeyar.com