Matt Ragoza

PhD student integrating deep learning with physics to solve important problems in science and medicine. | @ISPPitt | Advised by @kbatmang

[Matt Ragoza @ragozam]

@DaddyoShabazz @solmonkey79 @dzviovich @nntaleb assuming the field of view is the same in each image, but it's not

[Matt Ragoza @ragozam]

@Brad08414464 @getjonwithit Physics is the same everywhere and everytime

[Matt Ragoza @ragozam]

I was thrilled to present my latest work "Physics-Informed Neural Networks for Tissue Elasticity Reconstruction in Magnetic Resonance Elastography" last week at #MICCAI2023.

Video: https://t.co/MiujVZp9fI
Paper: https://t.co/lI4ApSLUWI
Code: https://t.co/NIFI8oir9h

[Matt Ragoza @ragozam]

@MelMitchell1 this seems to miss the forest for the trees

[Matt Ragoza @ragozam]

@francoisfleuret @visarga jpeg compression also parameterizes images in a manner very different from input samples

[Matt Ragoza @ragozam]

@awesomeintheory @MilitantHobo @JanelleCShane Exactly. If a human has to cherry pick/prompt engineer to get "knowledge" out of the model, its not knowledge, it's guessing

[Matt Ragoza @ragozam]

@bradleybuda @johndburger @DicksonPau @paulg they don't have to be right angles, the sides just have to be parallel

[Matt Ragoza @ragozam]

@Areyou52 @GDurMugen @Beck_Laboratory @demishassabis @jankosinski @HummerLab @MPIbp @embl This is a large complex of dozens of different proteins, which is much more impressive to simulate. It's the structure that allows messages to pass into and out of the cell nuclear to allow gene transcription and cellular reproduction

[Matt Ragoza @ragozam]

@fchollet is the handwriting style not considered information?

[Matt Ragoza @ragozam]

@littleidiotgirI how do you fail a class 3 times in 2 years?

[Matt Ragoza @ragozam]

@TheAgeofShoddy you realize he's mocking a common right wing talking point about what it means for something to be voluntary, right?

[Matt Ragoza @ragozam]

@Virtual1nstinct @macodiseas havent heard a definition of general intelligence that doesn't either 1) reference actual human abilities or 2) invoke entirely hypothetical godlike powers, and in any case you'd have to pass through 1) to get to 2)

[Matt Ragoza @ragozam]

@Virtual1nstinct someone in that previous civ probably said the same thing

[Matt Ragoza @ragozam]

@surface_model @Ayjchan https://t.co/kpIO5eqDdy

[Matt Ragoza @ragozam]

@JFPuget @balazskegl But I order for it to be realized and measured as some type of intelligence it has to be spelled out. The internal model of GPT3 probably has little resemblance to language.

[Matt Ragoza @ragozam]

@balazskegl Bad argument, you could have trained an ngram model on the same data it would be nowhere nearly as intelligent as GPT-3. Amount of data isn't the only factor.

[Matt Ragoza @ragozam]

@curdledcummies @viridator @empathy2000 @4HourL that's basically what being "under insured" is

[Matt Ragoza @ragozam]

@BenSpielberg its the "we need to support moderates so that we have the votes to pass bold legislation" but "we can't push for bold legislation that would make us lose moderate seats" snake eating its tail