New preprint (link ↓): Connectionists think that plasticity in the brain might enable it to learn like an artificial neural network. Machine learning people have some ideas about learning, and neuroscientists have some ideas about plasticity. (1/N)

@DBashIdeas The yeshivish obsession with sportsball is bizarre. Honestly one of the worst parts of Jewish education for boys. It's like the one acceptable outlet for extra-religious hobbies, and if you heaven forfend are interested in anything else, you're out of luck.

@burkov This is not true. Most major LLMs are pretty good at one-shot from novel input. I use this feature hundreds of times a day when coding with Cursor.

@TonyZador Well it definitely is a "move fast and break things" approach.

@TonyZador In some cases the overhead costs may make sense, in others they might not. (And this isn't just a question of the NIH, it's academia generally). I think it's reasonable for there to be tighter controls on whether public funds are just being used to top off university coffers.

@TonyZador Now tbf these universities do also have large operating costs, but a substantial amount of this cost is bloat, and also tuition covers some of it; tuition is definitely more expensive than just teaching fees.

@TonyZador I don't really know what 'good faith' means in this context. A lot of the major universities already have insanely large endowments (that also have large returns), so why is the government giving additional money? Give it directly to the researchers.

@TonyZador Like the important thing isn't that everything is exactly calculated for each individual lab, what's important is that there isn't some dark money left over that just goes to the university without being spent on actual research.

RT @Corca_math: We just launched Corca—a collaborative math editor. The main idea: there’s no 'text editor' for math. If you want to work o…

RT @eLife: The first complete 3D reconstruction and morphofunctional mapping of an insect eye.

@csaez_math Even theoretically, numbers that you can think of are describable numbers, are there theorems about the ratio of rational numbers to describable numbers?

@kaseyklimes They're an investment with a return is social status, not money.

RT @dav1dcg: A synapse-specific refractory period for plasticity at individual dendritic spines Potentiation at a single synapse is suffic…

@erikphoel It's also a pretty powerful tool for socialization and cultural transmission. E.g. to the extent that American culture is globalized, it's largely because of Hollywood.

@incelianitzen @ben_golub Econ 101 is the rigorous fundamentals.

RT @farznaj: Our paper is out: How Reliably Do Neurons Transmit Information? Computational modeling of Allen Cell Types dataset reveals h…

RT @IdoAizenbud: Human cortical pyramidal neurons are larger, exhibit more complex branching patterns, and feature distinctly nonlinear bio…