Profile avatar
marinagirl.bsky.social
pilates and vibes 🪼⛵✡️
45 posts 139 followers 1,722 following
Regular Contributor
Active Commenter

Here's an example of using #clojure to define a data model for CAD geometry and operations. From that model, I can create a STEP file via #FreeCAD and also view/export a meshed solid via #OpenSCAD

we met by the Bruhat-Tits tree

Mutation doesn't break referential transparency as long as it preserves equality. This is quite useful if you have fixpoints or HoTT.

It's done! After *checks notes* 6 months of yak shaving, I finally finished Part 3 of my Pragmatic Category Theory series! In this part, I give three more real-world examples of applying Semigroup while exploring why Associativity truly matters in detail. chshersh.com/blog/2024-12...

Functional Programming self-affirmations: 1. Parse, Don’t Validate 2. Make Illegal States Unrepresentable 3. Errors as values 4. Functional Core, Imperative Shell 5. Smart Constructor Repeat daily in front of a mirror for 2 minutes.

made the world's worst turing machine inside @tldraw.com computer

birbs

This paper on the usefulness of diversity of neuron types is pretty cool: journals.plos.org/ploscompbiol...

Have you ever wondered why diffusion models memorize and all initializations lead to the same training sample? As we show, this is because like in dynamic systems, the memorized sample acts as an attractor and a corresponding attraction basin is formed in the denoising trajectory.

Outstanding presentation, finally! DynaMo: In-Domain Dynamics Pretraining for Visuo-Motor Control @jeffacce.bsky.social @lerrelpinto.com

We now publish rich fMRI data in Scientific Data where participants viewed magic trick videos. Curiosity, memory, and reward manipulation are available: www.nature.com/articles/s41... This is a result of incredible effort by Stef Meliss (first author). Congrats!!

MBE - Course materials for Modern Binary Exploitation by RPISEC : github.com/RPISEC/MBE

🧵 At NeurIPS last year, we introduced Sequential Attachment-based Fragment Embeddings (SAFE), a novel line notation for chemical structures that reimagines SMILES strings as an unordered sequence of interconnected fragment blocks. youtu.be/oJsj5vWmD3c

Happy to see this work led by Zaid Zada now published in Neuron! We use LLM embeddings to capture word-by-word linguistic content transmitted from the speaker's brain to the listener's brain in real-time, face-to-face conversations: www.cell.com/neuron/fullt...

Proud of my first contribution to memory research: “Brain and eye movement dynamics track the transition from learning to memory-guided action” out in @currentbiology.bsky.social, great team effort together with Janina Klingspohr, Marcel Kehl & Bernhard Staresina. www.cell.com/current-biol...

Naturalistic encoding of concepts in the brain @viktorkewenig.bsky.social shows that, while concepts generally encode habitual experiences, the underlying neurobiological organisation is not fixed but depends dynamically on available contextual information. 👏🍾🐐 elifesciences.org/articles/91522

New paper story time (now out in PNAS)! We developed a method that caused people to learn new categories of visual objects, not by teaching them what the categories were, but by changing how their brains worked when they looked at individual objects in those categories. www.pnas.org/doi/10.1073/...

New paper from our lab, by Mark Saddler, using machine learning to test the role of temporal coding in hearing. Here is a quick summary. (1/n) www.nature.com/articles/s41...

The 25m Dwingeloo radio telescope (owned by ASTRON) has been operating since '50s. Recently, it has been used more by amateurs and pro-am radio astronomers. Excitingly, they picked up the Voyager 1 signal. Voyager 1 is ~25 billion km away! (~4 x Pluto distance) www.camras.nl/en/blog/2024... 🔭📡🧪

Vesta from Dawn on Aug 20, 2011. Approx true color image and a false color one. The true color image is from red, green and blue images with the overall color corrected using Vesta's spectrum. The false color is from 918, 749 and 438 nm displayed as RGB. Each image is a mosaic of 17 color frames.

Foresight is 2024