3 ms·

I really wish this book would include probability and statistic sections. My guess is that a lot of people, like me, will want to read it because they're gettin

I really wish this book would include probability and statistic sections. My guess is that a lot of people, like me, will want to read it because they're getting started with ML and need help getting used to the math, and probability/stats is an important part of it that's missing. Any chance you could add it in the future?

j2kun 8y ago
Well, two chapters have singular value decomposition and neural networks as the applications. So it does have a lot of ML :) But yes, I unfortunately had to cut a probability chapter. I think someone who reads this book would have a much easier time learning probability after, and a better foundation.
- yantrams 8y ago
  Have just ordered the e-book. Didn't expect to find a chapter on Hyperbolic tessellations. Nice!
- bsg75 8y ago
  > I think someone who reads this book would have a much easier time learning probability after, and a better foundation. So will you be following up with a prob & stats book? ;)
- codesushi42 8y ago
  Thank you for this book. There is a huge need for this. It saddens me that schools may be handing out CS degrees without having first required students to at least have taken linear algebra, multivariable calculus, and discrete math that covers basic counting, sets, graphs and groups. How can this be? Probability and statistics should also be a required part of every CS program.
  - j2kun 8y ago
    /shrug misaligned incentives probably. Schools are also handing out CS degrees without students being all that good at writing programs either.
melbourne_mat 8y ago
https://greenteapress.com/wp/think-stats-2e/ https://greenteapress.com/wp/think-stats-2e/
codesushi42 8y ago
Unlike the other subjects contained in this book, I do not think that any useful understanding of Probability and Statistics could result from only one or two chapters of coverage. I would recommend a book I found that covers Statistics using Python: Statistical Methods for Machine Learning https://machinelearningmastery.com/products/ https://machinelearningmastery.com/products/ Personally, I much prefer that book to the free "Think Stats" one available. You can also find a book there on Linear Algebra and others related to Deep Learning, all containing real implementations with Python. I have no affiliation with the site; these books have proven so valuable to me that I need to share them. Anyway, the book mentioned in the OP looks like a great foundation and I picked up a copy as a refresher. I would have also liked to have seen coverage of number theory in this book, with RSA being an application, but that too would have required lengthy explanation for any useful insights to be learned.
madhadron 8y ago
That's rather like saying that a Java book should include a section on processor architecture so you can design your own chips. Plus, what's in this book isn't really enough to do probability and statistics at any but the most basic level. The path to a deep understanding of inference that makes most modern ML seem straightforward goes through measure theory, Lebesgue integration, some functional analysis, a smattering of topology, and an intuition around dynamical systems. Then you go do probability and random processes, detour into game theory for a bit, learn decision theory, and at that point it all goes from mystical to bleedin' obvious.

j2kun 8y ago

Well, two chapters have singular value decomposition and neural networks as the applications. So it does have a lot of ML :) But yes, I unfortunately had to cut a probability chapter. I think someone who reads this book would have a much easier time learning probability after, and a better foundation.

yantrams 8y ago
Have just ordered the e-book. Didn't expect to find a chapter on Hyperbolic tessellations. Nice!
bsg75 8y ago
> I think someone who reads this book would have a much easier time learning probability after, and a better foundation. So will you be following up with a prob & stats book? ;)
codesushi42 8y ago
Thank you for this book. There is a huge need for this. It saddens me that schools may be handing out CS degrees without having first required students to at least have taken linear algebra, multivariable calculus, and discrete math that covers basic counting, sets, graphs and groups. How can this be? Probability and statistics should also be a required part of every CS program.
- j2kun 8y ago
  /shrug misaligned incentives probably. Schools are also handing out CS degrees without students being all that good at writing programs either.

melbourne_mat 8y ago

https://greenteapress.com/wp/think-stats-2e/ https://greenteapress.com/wp/think-stats-2e/

codesushi42 8y ago

Unlike the other subjects contained in this book, I do not think that any useful understanding of Probability and Statistics could result from only one or two chapters of coverage. I would recommend a book I found that covers Statistics using Python: Statistical Methods for Machine Learning https://machinelearningmastery.com/products/ https://machinelearningmastery.com/products/ Personally, I much prefer that book to the free "Think Stats" one available. You can also find a book there on Linear Algebra and others related to Deep Learning, all containing real implementations with Python. I have no affiliation with the site; these books have proven so valuable to me that I need to share them. Anyway, the book mentioned in the OP looks like a great foundation and I picked up a copy as a refresher. I would have also liked to have seen coverage of number theory in this book, with RSA being an application, but that too would have required lengthy explanation for any useful insights to be learned.

madhadron 8y ago

That's rather like saying that a Java book should include a section on processor architecture so you can design your own chips. Plus, what's in this book isn't really enough to do probability and statistics at any but the most basic level. The path to a deep understanding of inference that makes most modern ML seem straightforward goes through measure theory, Lebesgue integration, some functional analysis, a smattering of topology, and an intuition around dynamical systems. Then you go do probability and random processes, detour into game theory for a bit, learn decision theory, and at that point it all goes from mystical to bleedin' obvious.