should i do some self-study or take a masters to break into ML? ive always been more interested in the research side of things rather than the productionalization of software, but i dont want to get stuck in academia. i also really enjoy working with math (even tho stats is bs, thats a diff story tho). I dont mind becoming an MLE either but i dont know how to break in from SWE. maybe its startup time?
ML is a subject of stats and CS. If you think stats is all about confidence intervals I agree with Google.
Of course it isn’t. I enjoy the math part of stats, creating a model, fitting the data, etc. I don’t like traditional stats is what I meant, the basic stuff. It’s misused by many ‘scientists’ these days to produce ridiculous claims in an attempt to sway public option.
My point, if it was missed, is that you will be even more disappointed by blacker-box ML models and the culture in CS trained practitioners. If you truly studies statistics from the theoretical standpoint, you will be sorely frustrated by the current state of ML in industry. Turn back now - I say this with all earnestness.
Don't worry. From your comments above I am sure you won't get stuck in academia.
Thanks neutron
Wayfair - I believe the comments in this thread are coming from a good place. It would be worth your while to spend some time to learn about the "stats bs" - seriously.
Ok, thank you, I appreciate it. I am currently studying Andrew Ng’s course. Maybe I phrased it badly, I meant that I don’t want to be like one of the fake ass scientists who spew out bullshit studies in the name of science and use flawed statistical models to assert their confidence in their results.
If you think stats is BS, you're probably very far from being qualified for any serious ML work. I'd guess that you in particular need at least a masters, or equivalent online education (which should include plenty of statistics)
I’ve studied stats... up to SVM, time series analysis, some basic neural nets, pca, and as well as the quantum model, stat therm models (Boltzmann, Einstein, etc). The math part of them is quite intuitive. The stats not so much. I can do statistics. How stats is being used is bullshit.
If you want to compare universities based on their ML research publications, you can check this link: http://csrankings.org/#/index?mlmining
For those who don't want to bother opening the link.
Thank you! CMU has always been a goal of mine, but I’ve been eyeing MIT since I’m in boston now. Does MIT have a good part time/online program?
It depends on how good you are. The more you can accomplish on your own, the less you need formal degrees. Obviously, given how crowded the field is now, it's harder to stand out, so you'll have to make that call
If you don't want to get a PhD get ready to get rejected for MLE roles. It's tough out there rn
India
13h
1161
Lost respect for Modiji
2024 Tax
12h
1982
Biden’s new tax proposal is wild
Tech Industry
18h
2509
Go woke, go broke: Google fires 28 employees involved in pro-Hamas protest
2024 Presidential Election
4h
1101
Biden ruined America and tech! Tax plans are insane
Personal Finance
Yesterday
3182
IRS took 73K 😭
Irony of calling stats bs when you want to do ML
Go ahead and defend it all you want. defining a random 95% confidence interval and fitting models because of muh feelings is 100% bullshit. The linear algebra, matrix calculus, etc. however, are not.
That only shows you have little formal training beyond probably some 100-level undergrad course at a tier 2 school