Science Memes

11021 readers

3990 users here now

Welcome to c/science_memes @ Mander.xyz!

A place for majestic STEMLORD peacocking, as well as memes about the realities of working in a lab.

Rules

Don't throw mud. Behave like an intellectual and remember the human.
Keep it rooted (on topic).
No spam.
Infographics welcome, get schooled.

This is a science community. We use the Dawkins definition of meme.

Research Committee

[email protected]

Other Mander Communities

Science and Research

Biology and Life Sciences

Physical Sciences

Humanities and Social Sciences

Practical and Applied Sciences

Memes

Miscellaneous

founded 2 years ago

MODERATORS

[email protected]

528

answer = sum(n) / len(n) (lemmy.eco.br)

submitted 4 months ago by [email protected] to c/[email protected]

119 comments fedilink hide all child comments

you are viewing a single comment's thread
view the rest of the comments

[–] [email protected] 3 points 4 months ago (1 children)

That's where the almost comes in. Unfortunately, there are many traps for the unwary stochastic parrot.

Training a neural net can be seen as a generalized regression analysis. But that's not where it comes from. Inspiration comes mainly from biology, and also from physics. It's not a result of developing better statistics. Training algorithms, like Backprop, were developed for the purpose. It's not something that the pioneers could look up in a stats textbook. This is why the terminology is different. Where the same terms are used, they don't mean quite the same thing, unfortunately.

Many developments crucial for LLMs have no counterpart in statistics, like fine-tuning, RLHF, or self-attention. Conversely, what you typically want from a regression - such as neatly interpretable parameters with error bars - is conspicuously absent in ANNs.

Any ideas you have formed about LLMs, based on the understanding that they are just statistics, are very likely wrong.

[–] [email protected] 2 points 4 months ago (1 children)

"such as neatly interpretable parameters"

Hahaha, hahahahahaha.

Hahahahaha.

[–] [email protected] 0 points 4 months ago (1 children)

If parameters aren’t neatly interpretable then it’s bad statistics. You’ve learned nothing about the general structure of the data.

Linear regression models are often great tools for explaining the structure of the data. You can directly see which parts of the input are more important for determining the output. You have very little of that when using neural networks with more than 1 hidden layer.

[–] [email protected] 0 points 4 months ago (1 children)

"If parameters aren’t neatly interpretable then it’s bad statistics."

Haha, keep going guys. You obviously know a lot about statistics.

[–] [email protected] 0 points 4 months ago (1 children)

https://www.nature.com/articles/nmeth.4642

This article use different wording than me, but in essence: Statistics is mostly about using a known model to explain the data. Machine Learning is mostly about finding any model that predicts the data well. Different purposes with some overlap. Some statistical methods are used in Machine Learning, but that doesn’t necessarily mean all of Machine Learning is statistics.

The boundary between statistical inference and ML is subject to debate—some methods fall squarely into one or the other domain, but many are used in both. […] Statistics requires us to choose a model that incorporates our knowledge of the system, and ML requires us to choose a predictive algorithm by relying on its empirical capabilities.

Another (potentially lower quality) article that is not from Nature, but discusses the meme in particular:

https://www.datarobot.com/blog/statistics-and-machine-learning-whats-the-difference/

Because of the large number of variables in machine learning datasets, the models developed from them can be simultaneously extremely accurate and almost impossible to understand. Statistical models, on the other hand are typically easier to understand because they are based on fewer variables, and the accuracy of relationships is supported by tests of statistical significance.

[–] [email protected] 1 points 4 months ago* (last edited 4 months ago)

Seeing your comment I wondered how someone publishing in Nature could have possibly left out the use of statistics for prediction. That would be a wild oversight that only someone with little knowledge of the topic would make, and surely not something that the editors of Nature would miss. Upon clicking the link I see that they mentioned it in the very first sentence and apparently ignore it if someone happens to call the prediction model a machine learning model. Using statistical models for prediction has been used since the start of the field, and renaming things that have been used for decades as "machine learning" doesn't suddenly make them not statistics.

Artificial neural networks are statistical models, with numerous statistical approaches associated with their use, development and interpretation.