Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Understanding Machine Learning: From Theory to Algorithms (2014)

cs.huji.ac.il

Post reply on HN
Print view View original

11–20 of 58 posts

Re: Understanding Machine Learning: From Theory to Algorithms (2014)

#11

Post by alfiedotwtf »

cdnsteve wrote: post #7

I feel like the barrier to machine learning for me, as I've seen in many tutorials and books and is an immediate discouragement, is the massive amount of math thrown in your face. Many of us didn't just graduate, need glasses and fall asleep at 8pm on the couch when the kids are in bed... Math is this distant fragment of memory buried under years of everything not Math. It feels like machine learning is only taught b…

The problem is that machine learning is applied probability and statistics.

If you're interested, here's a gentle book that should give you enough background:

http://www.amazon.com/Modern-Introduction-Probability-Statis...

Re: Understanding Machine Learning: From Theory to Algorithms (2014)

#12

Post by davmre »

cdnsteve wrote: post #7

I feel like the barrier to machine learning for me, as I've seen in many tutorials and books and is an immediate discouragement, is the massive amount of math thrown in your face. Many of us didn't just graduate, need glasses and fall asleep at 8pm on the couch when the kids are in bed... Math is this distant fragment of memory buried under years of everything not Math. It feels like machine learning is only taught b…

Andrew Ng's Coursera ML course is supposed to be pretty accessible. I've also heard good things about Machine Learning for Hackers (http://www.amazon.com/Machine-Learning-Hackers-Drew-Conway/d...).

Ultimately, ML is a mathematical discipline. You can ask for a gentle approach that gets you to the foot of the mountain, but "if you want to learn about nature, to appreciate nature, it is necessary to understand the language that she speaks in." If you want to be more than an amateur, there's not much substitute for getting comfortable with math at the level of, say, Kevin Murphy's book.

The good news is that the required math is fairly elementary - calculus, linear algebra, probability and statistics, all freshmen or maybe sophomore-level topics - so it shouldn't be beyond reach of a motivated developer able to set aside some time to learn. MOOCs and organizing study groups with friends/co-workers can help a lot here as well.

Re: Understanding Machine Learning: From Theory to Algorithms (2014)

#13

Post by zintinio4 »

alvern wrote: post #10

My question for someone that has an intermediate level of skill in machine learning, what's the best way to dip your toes in? (Udacity, coursera, edx, PDFs, Talking Machines podcast, etc)

Find a problem to work on in a domain you find interesting. By reading published papers and trying to attack the problem, you'll be forced to pick up a lot of other knowledge not commonly discussed like: feature extraction and selection, dimensionality reduction, dealing with sparsity, common metrics for that problem, recent work, etc.

I was forced to learn a massive amount in a short period of time for work, but I'd previously watched Andrew Ng's lectures, as well as majored in Math/CS. I can also generally recommend Hinton's NN lectures, Socher's Deep learning for NLP, Andrew Ng's Machine Learning, and a few books.

Re: Understanding Machine Learning: From Theory to Algorithms (2014)

#14

Post by dtjones »

cdnsteve wrote: post #7

I feel like the barrier to machine learning for me, as I've seen in many tutorials and books and is an immediate discouragement, is the massive amount of math thrown in your face. Many of us didn't just graduate, need glasses and fall asleep at 8pm on the couch when the kids are in bed... Math is this distant fragment of memory buried under years of everything not Math. It feels like machine learning is only taught b…

Applied machine learning without understanding of the fundamental mathematical assumptions can be a recipe for failure

That said, there are some graphical examples to help understand of how learning algorithms work in 2 dimensions.

Re: Understanding Machine Learning: From Theory to Algorithms (2014)

#15

Post by fapjacks »

davmre wrote: post #12
cdnsteve wrote: post #7

I feel like the barrier to machine learning for me, as I've seen in many tutorials and books and is an immediate discouragement, is the massive amount of math thrown in your face. Many of us didn't just graduate, need glasses and fall asleep at 8pm on the couch when the kids are in bed... Math is this distant fragment of memory buried under years of everything not Math. It feels like machine learning is only taught b…

Andrew Ng's Coursera ML course is supposed to be pretty accessible. I've also heard good things about Machine Learning for Hackers ( http://www.amazon.com/Machine-Learning-Hackers-Drew-Conway/d... ). Ultimately, ML is a mathematical discipline. You can ask for a gentle approach that gets you to the foot of the mountain, but "if you want to learn about nature, to appreciate nature, it is necessary to understand the la…

Yeah I'd like to second this. This course specifically is what opened the door for me.

Re: Understanding Machine Learning: From Theory to Algorithms (2014)

#17

Post by subnaught »

davmre wrote: post #12
cdnsteve wrote: post #7

I feel like the barrier to machine learning for me, as I've seen in many tutorials and books and is an immediate discouragement, is the massive amount of math thrown in your face. Many of us didn't just graduate, need glasses and fall asleep at 8pm on the couch when the kids are in bed... Math is this distant fragment of memory buried under years of everything not Math. It feels like machine learning is only taught b…

Andrew Ng's Coursera ML course is supposed to be pretty accessible. I've also heard good things about Machine Learning for Hackers ( http://www.amazon.com/Machine-Learning-Hackers-Drew-Conway/d... ). Ultimately, ML is a mathematical discipline. You can ask for a gentle approach that gets you to the foot of the mountain, but "if you want to learn about nature, to appreciate nature, it is necessary to understand the la…

I'm currently taking Andrew Ng's Coursera course and I'd agree it's quite accessible. In fact, if you have a solid understanding of calculus and linear algebra, you might find it a bit slow at times.

Re: Understanding Machine Learning: From Theory to Algorithms (2014)

#18

Post by ska »

cdnsteve wrote: post #7

I feel like the barrier to machine learning for me, as I've seen in many tutorials and books and is an immediate discouragement, is the massive amount of math thrown in your face. Many of us didn't just graduate, need glasses and fall asleep at 8pm on the couch when the kids are in bed... Math is this distant fragment of memory buried under years of everything not Math. It feels like machine learning is only taught b…

Machine learning is fundamentally mathematical, so you can't expect to completely avoid it without remembering or learning at least a little bit of the maths. Trying to avoid it all won't do you any favors, it will just mean you can't understand what is going on.

On the other hand, the math you absolutely need to follow along is pretty straightforward, so you can hopefully find tutorials that emphasize the applications and graphic examples of what is going on.

Re: Understanding Machine Learning: From Theory to Algorithms (2014)

#19

Post by visarga »

cdnsteve wrote: post #7

I feel like the barrier to machine learning for me, as I've seen in many tutorials and books and is an immediate discouragement, is the massive amount of math thrown in your face. Many of us didn't just graduate, need glasses and fall asleep at 8pm on the couch when the kids are in bed... Math is this distant fragment of memory buried under years of everything not Math. It feels like machine learning is only taught b…

> the massive amount of math thrown in your face

I can skip the math part - we don't all invent new algorithms - but what I really need is a large enough & gradual set of problems to solve (datasets + verification scripts). I mean, start from the simplest and teach people how to use the already available software. Machine Learning should be assimilated practically, too much theory with too little application is useless. Most of us should focus on using existing software efficiently instead of being able to implement backprop.

Re: Understanding Machine Learning: From Theory to Algorithms (2014)

#20

Post by spdustin »

visarga wrote: post #19
cdnsteve wrote: post #7

I feel like the barrier to machine learning for me, as I've seen in many tutorials and books and is an immediate discouragement, is the massive amount of math thrown in your face. Many of us didn't just graduate, need glasses and fall asleep at 8pm on the couch when the kids are in bed... Math is this distant fragment of memory buried under years of everything not Math. It feels like machine learning is only taught b…

> the massive amount of math thrown in your face I can skip the math part - we don't all invent new algorithms - but what I really need is a large enough & gradual set of problems to solve (datasets + verification scripts). I mean, start from the simplest and teach people how to use the already available software. Machine Learning should be assimilated practically, too much theory with too little application is usele…

Kaggle's "Titanic: Machine Learning from Disaster" [0] is a great place to start. If you're anxious to dive in, go straight to "Getting Started with Excel" [1] and move on to "Getting Started with Python" [2]

Note: I'd strongly recommend the Anaconda Python distribution [3], as it has pretty much everything you need. Also, for immediate feedback on what you're doing with Python, I've fallen in love with Jupyter Notebooks (formerly IPython Notebooks) [4], which you'll have as part of the Anaconda distribution, along with all the other popular Python packages for scientific work.

[0] https://www.kaggle.com/c/titanic

[1] https://www.kaggle.com/c/titanic/details/getting-started-wit...

[2] https://www.kaggle.com/c/titanic/details/getting-started-wit...

[3] https://www.continuum.io/downloads

[4] http://ipython.org

Post reply on HN