Excited to share our new @PNAS paper: We introduce “infomorphic” neurons that learn by maximizing individual, information-theoretic objectives. Our framework on interpretable neural learning is applicable across different learning paradigms. Read more: https://t.co/4pEwlAjMgy 1/2
Also, check out our ICLR’25 Oral Paper that extends this framework to larger networks and shows how to systematically identify and improve these per-neuron objectives: https://t.co/QX9HVnB6uk #neuralnetworks#infotheory#AI#ML#MachineLearning#locallearning 2/2
Excited to share our new @PNAS paper: We introduce “infomorphic” neurons that learn by maximizing individual, information-theoretic objectives. Our framework on interpretable neural learning is applicable across different learning paradigms. Read more: https://t.co/4pEwlAjMgy 1/2
We are organizing a focus session on "Self-regulating and learning systems: from neural to social networks" at the DPG Frühjahrstagung (Spring meeting of the German Physical Society from March 16-21, 2025 in Regensburg).
Abstract Deadline: 01.12.2024
https://t.co/SXdEg08URc
If you're interested in our take, have a look at section 4.3 (Fig. 4) in "A General Framework for Interpretable Neural Learning based on Local Information-Theoretic Goal Functions" https://t.co/XE1KNQTKt5
Neural Networks = Physics, end of discussion.. ;) Great stuff, congrats, and cool to let these learn via information theory: https://t.co/XE1KNQTKt5
Have a look here for a good introduction (Hopfield,Boltzmann):
https://t.co/oFS17GNrmc
https://t.co/NmXKpg3t3t
#NobelPrize2024
@ElsaArcaute My tip: make it interesting by interesting examples and mini-projects. Not by gamifying the language, as sometimes tried for kids. I was always confused how Robot Karol was supposed to be 'real programming'. 15 year persons are fast in getting the real thing if explained well
@SallyKamille@elaifresh I also lived there for half a year. And to me it is just the same as you described. Additionally, seasons are always motivating to go out and use the special times of the year for certain things - which did not really apply there, with the exception of wind
Solar has continuously fallen in price following scaling laws for half a century.
I'm sorry. I just don't buy that anybody can look at the data and in good faith conclude "solar is all subsidies"
Tell me. What has a bigger impact?
- a 30% tax credit in one country
- the 19 times prices fell by 30% in the last 50 years
Globally, across decades, different countries leading manufacturing, subsidies popping up and going away, etc. solar costs have steadily followed scaling laws.
How much of the 99.9% drop in cost do you attribute to these "secret subsidies"?
Is it all some global conspiracy to make solar look cheap while people actually lose thousands of dollars on every panel?
These arguments were absurd in the 90's. Today?? Give me a break.
@piet_privat Wir haben seit Jahren diese hier. Die Beschichtung ist nicht aufgeklebt und hält sehr lange. Kann sogar erneuert werden https://t.co/NODzoN8ZrJ
@oldbooksguy Agree with most of the points, and would add that consumption without clear intention (scrolling, binge-anything) costs a lot of time and energy that we would otherwise have for creating