Ask HN: How do I go about learning machine learning?

https://news.ycombinator.com/item?id=9510843

by zach_d • 11 years ago

25 13 11 years ago

news.ycombinator.com

I've finally found an interesting field that gets me really excited to program! I learned python a month ago but haven't had any interesting ideas but I just found out about Machine learning, AI, and data science. Really interesting stuff so I'm wondering what HN would suggest as a general learning path! What math should I know to really understand this stuff? What books do I need to read? Some beginner project ideas? Anything at all! I'm a HS Senior and I plan to really focus on machine learning, data science, and AI once I'm in college.

Thank you.

stonemetal• 11 years ago

Machine learning is applied statistics. Elements of Statistical Learning was written by profs at Stanford and is free on the internet. It might be a bit hard since they don't really teach stats in high school. O'Reilly has a few ml books. I liked Programming Collective Intelligence it is beginner friendly and would probably help you come up with project ideas.

LukeFitzpatrick• 11 years ago

This is one possible resource for you www.codecloud.me - what it does, people can code together on projects. The focus is on learning and unfortunately no payment. They do have some machine learning projects on there which could help you to start off & it comes with mentors.

playing_colours• 11 years ago

You can try "Data Science from Scratch" [0] to get some taste. It uses Python to teach essentials of data science, and ML altorithms. The code quality is very good, and there is an introduction to Statistics, Maths and Python to start.

Then you can continue with improving your maths (Linear Algebra [1], Calculus [2], [3]) and moving on with Statistical Learning [4] [5]. I am personally going now through this plan.

[0] http://shop.oreilly.com/product/0636920033400.do

[1] http://www.amazon.com/Linear-Algebra-Right-Undergraduate-Mat...

[2] http://www.amazon.com/Calculus-4th-Michael-Spivak/dp/0914098...

[3] http://www.amazon.com/Calculus-Manifolds-Approach-Classical-...

[4] http://www-bcf.usc.edu/~gareth/ISL/

[5] http://statweb.stanford.edu/~tibs/ElemStatLearn/

misframer• 11 years ago

An Introduction to Statistical Learning [0] is also good. It's a little less technical than The Elements of Statistical Learning. We used it for our statistical learning course at my university. The full PDF is available for free as well [1].

[0] http://www-bcf.usc.edu/~gareth/ISL/

[1] http://www-bcf.usc.edu/~gareth/ISL/ISLR%20Fourth%20Printing....

lovelearning• 11 years ago

I recommend starting with Andrew Ng's ML course on Coursera. He teaches and presents the subject in a way that is both fun and non-intimidating to a beginner. You get a good overview of techniques, and get to try out fun exercises like digit recognition.

yen223• 11 years ago

Get your fundamentals down. Focus heavily on linear algebra and probability theory.

If you think you've nailed Python, go ahead and look at NumPy and SciPy - they're different enough from Python, and they crop up very often in ML.

bra-ket• 11 years ago

https://www.quora.com/How-do-I-learn-machine-learning-1

moserware• 11 years ago

See previous discussion at https://news.ycombinator.com/item?id=9432952 . My advice in particular is at https://news.ycombinator.com/item?id=9433981

murbard2• 11 years ago

A lot of the advice you will get is going to be in the spirit of the American way of teaching various scientific disciplines. I would decribe it as a hands on approach, that gets you started really fast and doing exciting things. This is true of computer science, statistics, and physics as far as I know. It's probably a good way to go about learning these things.

The French method for teaching these things (the one I'm used to) is to first do a lot of mathematics and then to introduce these topics through powerful mathematical formalisms, from the get go. All else equal, for most people, it's probably a worse approach.

However, the American way does err a little bit on the side of too little mathematics, and it can come and bite you. My suggestion would thus be to follow all the advice given, but give 50% more emphasis on mathematics than people would otherwise suggest. Fortunately, the mathematics of machine learning aren't too complicated, but you want to be able to breathe that stuff. In particular this means:

- linear algebra

- real and complex analysis

- multivariate calculus

- topology (just a bit)

- probability theory

It's like touch typing. It may be tedious at first, but once you know this stuff really well, learning new things will be a breeze.

eli_gottlieb• 11 years ago

I recommend Understanding Machine Learning for a textbook, as I rather enjoyed a class using it.

http://www.cambridge.org/al/academic/subjects/computer-scien...

KodrAus• 11 years ago

As someone who came into this field with only an average high-school-level grasp of mathematics I found that particular book incredibly worthwhile. It's taken me almost 12 months to start understanding it enough to actually apply it but it's worth it.

chollida1• 11 years ago

I wrote about this here: https://news.ycombinator.com/item?id=8767092 and here: https://news.ycombinator.com/item?id=9433316

Long story short, the biggest mistake I see people making is not actually rolling up their sleeves and learning the math.

People are often content to watch hour after hour of Udacity, Khan academy and Coursera videos but the applied follow up is where most people drop off. At the very least any course work should be followed up by something practical like a kaggle exercise to prove that you can apply the technique you just learned. Consider the benefit of just watching videos vs doing actual applied work.

On one hand if you just watch videos you might learn alot but how do you prove that to someone hiring you? On the other hand if you sit down and spend a week attaching a Kaggle excise then at the very least you have something to point people to, to show that you can apply machine learning techniques.

My recommendation has always been to read the first 5 chapters of Introduction to statistical learning: http://www-bcf.usc.edu/~gareth/ISL/

and if you fly thorugh it then sample Elements of statistical learning http://statweb.stanford.edu/~tibs/ElemStatLearn/ for the topics that you want to learn.

If intro to statistical learning is too advanced, then go to Khan academy and work your way through their statistics videos.

From my experience you can bucket people into skill level by looking at how they attack a new problem.

Beginners tend to start by saying they'll need a hadoop cluster and spend the next week setting up a pipeline.

Intermediate people tend to jump into R or scikit and try to model the problem with a small subset of data and the library and technique they know best.

The advanced people tend to flesh out their hypothesis first and then work out the math and then jump to modelling with a small set of data and finally move to a cluster.

rayalez• 11 years ago

Hi! I've compiled a list of the best resources that I'm aware of: http://digitalmind.io/post/deep-learning

Ask HN: How do I go about learning machine learning?

Related Stories