# ML Prediction & Scientific Inference

**URL:** <https://discuss.tinyml.seas.harvard.edu/t/ml-prediction-scientific-inference/834>\
**Category:** General Discussions\
**Created:** [August 7, 2021, 1:40pm UTC](https://discuss.tinyml.seas.harvard.edu/t/ml-prediction-scientific-inference/834 "2021-08-07T13:40:41Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![aquacalc](https://yyz1.discourse-cdn.com/flex027/user_avatar/discuss.tinyml.seas.harvard.edu/aquacalc/32/223_2.png) [@aquacalc](https://discuss.tinyml.seas.harvard.edu/u/aquacalc)\
**Post date:** [August 7, 2021, 1:40pm UTC](https://discuss.tinyml.seas.harvard.edu/t/ml-prediction-scientific-inference/834/1 "2021-08-07T13:40:41Z")

</div>

Some of my colleagues are unfamiliar with the ML paradigm, so I wrote a short summary that introduces the basics and explains how I plan to apply ML in our domain. It draws in large part on what I learned in Vijay’s courses, which did a lot to crystalize my ideas.

I’ll pass along an article I came across that frames the distinction between traditional scientific models and ML in a compelling way:

[_“New machine learning theory raises questions about [the] nature of science”_](https://www.sciencedaily.com/releases/2021/02/210212094120.htm)

(I would have removed the word “New” from the title.)

The ‘money’ quote – and likely a controversial assertion – that jumped out at me:

> “I would argue that the ultimate goal of any scientist is prediction,” [Princeton physicist Dr. Hong] Qin said. "You might not necessarily need a law. For example, if I can perfectly predict a planetary orbit, [then] I don’t need to know Newton’s laws of gravitation and motion.

Within the context of sustainable aquaculture, my interpretation is [_here_](https://peaceful-fjord-26178.herokuapp.com/about#qin). (NB: It’s on Heroku’s free tier, so it doesn’t open immediately.)