3 ms·

When people talk about the growth (or sometimes 'excess') of ML solutions these days, I always wonder about this. A basic linear regression probably isn't ML,

When people talk about the growth (or sometimes 'excess') of ML solutions these days, I always wonder about this. A basic linear regression probably isn't ML, a backprop neural net clearly is, but somewhere between the two is a very fuzzy line between "statistics and data cleaning" and "actually machine learning". I think a lot of people have just pushed the ML angle of an already-reasonable approach to tie into that popularity.

yessql 10y ago
ML courses often start with linear regression, and if you build up complicated polynomials to find a nonintuitive model of your problem, I would definitely consider that machine learning.
- sidlls 10y ago
  I wouldn't. I'd call it "basic computational statistics." But I think I might be in the minority on that.
  - Bartweiss 10y ago
    I've always thought of regressions, even high-order ones, as just a statistical tool. They're present at the start of ML courses, sure, but as a tool used in ML techniques or a good alternative to them. It looks like that's not the standard view, though.
    - LolWolf 10y ago
      The whole deal seems weird to me. Neural networks are just functional approximators, so why isn't a linear regressor of k-th order (e.g. Taylor expansion up to k-th order) also considered "ML"? What's the distinction here?
BickNowstrom 10y ago
I think these distinctions are akin to discussing musical genre in Youtube comments. Something can be all of: mathematics, advanced IT, computer science, data science, statistics, econometrics, control theory, pattern recognition, machine learning, predictive modeling, data mining, data analysis, data engineering, or even AI if you are feeling fancy.
NumberCruncher 10y ago
I like this definition of ML. Off topic: somehow the people not doing data cleaning because it is so boring and ML techniques do not need clean data end up with the worst overfitting I have ever seen.

yessql 10y ago

ML courses often start with linear regression, and if you build up complicated polynomials to find a nonintuitive model of your problem, I would definitely consider that machine learning.

sidlls 10y ago
I wouldn't. I'd call it "basic computational statistics." But I think I might be in the minority on that.
- Bartweiss 10y ago
  I've always thought of regressions, even high-order ones, as just a statistical tool. They're present at the start of ML courses, sure, but as a tool used in ML techniques or a good alternative to them. It looks like that's not the standard view, though.
  - LolWolf 10y ago
    The whole deal seems weird to me. Neural networks are just functional approximators, so why isn't a linear regressor of k-th order (e.g. Taylor expansion up to k-th order) also considered "ML"? What's the distinction here?

BickNowstrom 10y ago

I think these distinctions are akin to discussing musical genre in Youtube comments. Something can be all of: mathematics, advanced IT, computer science, data science, statistics, econometrics, control theory, pattern recognition, machine learning, predictive modeling, data mining, data analysis, data engineering, or even AI if you are feeling fancy.

NumberCruncher 10y ago

I like this definition of ML. Off topic: somehow the people not doing data cleaning because it is so boring and ML techniques do not need clean data end up with the worst overfitting I have ever seen.