Great, it seems like our views are pretty harmonised.
Boost machine learning knowledge with Introduction to Statistical Learning and the associated Stanford online course. This is what my boss at work recommended to me. I’m not sure if it’s better than Andrew NG’s course.
Yeah, I read Introduction to Statistical Learning in R (ISLR) and then went on to Ng’s course. It was when I viewed the latter that it really clicked. I still think ISLR was good though. Ng uses Matlab (bad if you’re getting used to R), provides better explanations for the math (good), has quizzes (good) and focuses more on modern approaches including neural networks (good). So I would suggest at least enrolling in the Coursera course in order to have it available as a backup.
Re math prerequisites, Khan Academy is also how I got to being able to understand machine learning. I also did a Coursera course in Calculus, which might help. This Jim Fowler is a good teacher, similarly to Sal Khan.
To start learning machine learning, you’ll need basic linear algebra—how to multiply vectors and matrices basically. For the introductory stuff, you can apply it without understanding calculus, although it’s fundamental to how machine learning works and you’ll need to learn some calculus sooner or later. But if it helps with motivation, if you know the basics of how to multiply vectors and matrices, you can definitely start Ng’s course, and learn some of the rest in tandem. In the long-run, to get good at ML clearly requires proficiency with linear algebra, calculus and statistics, as well as a general comfort with math that seems to come from practice, so it seems like we have to put in a bunch of solid hours into getting these foundations. More hours per week seems good.
I’d love to visit SF for sure (I already meant to a few times, but never got around to it) and I would not be too surprised if I worked there at some point in the future. Right now, the job I have is in Chicago, though, and I think the benefits outweigh the costs for leaving my job at the moment, and I foresee that being true for at least the next six months.
I think that the several hundred dollars it costs you to make a flight there, and a week of lost salary would pay itself off in expected future earnings from the option of working there later, new professional contacts, new insights into cause prioritisation, fun, new friends, extra passion for doing good, etc. The new Global EA Summit will be coming up too, which might be a reason to get down there. You meeting people there just feels like all-round good news to me.
Yeah, I read Introduction to Statistical Learning in R (ISLR) and then went on to Ng’s course. It was when I viewed the latter that it really clicked.
Makes sense. I think I’ll try that in the same order as well.
Re math prerequisites, Khan Academy is also how I got to being able to understand machine learning. I also did a Coursera course in Calculus, which might help.
I don’t actually know the basics of multiplying vectors and matricies (I learned them in college but forgot soon afterward), so I should learn that first.
You’ve convinced me of two changes to make:
First, I should go in sequence with my learning rather than parallel. I think I’ll aim for Khan Academy Algebra → Khan Academy Calculus I → Khan Academy Calculus II → Khan Academy Linear Algebra → Introduction to Statistical Learning → Angrew Ng’s course. (I think I’ll still do Advanced R --> Learn Hadoop in parallel, though, because my R skills are somewhat unrelated to my MR skills.) (To-do for self: re-arrange learning list.)
Second, I should spend more than 2hrs/wk on this. I can probably cut out more EA time. (To-do for self: think on this more.)
-
The new Global EA Summit will be coming up too, which might be a reason to get down there. You meeting people there just feels like all-round good news to me.
Yeah, I’ll come out for the Global EA Summit and you’ve convinced me to try to make a full week of it. We have a pretty flexible vacation policy here, so I shouldn’t even lose salary. I just have to reconcile this with other vacation I plan on taking. (To-do for self: plan out vacation for 2015, watch for SF EA Summit dates.)
I’m thinking of also going to SF when Joey and Xio get around to visiting SF, but I don’t know if that’s going to be in 2015.
Great, it seems like our views are pretty harmonised.
Yeah, I read Introduction to Statistical Learning in R (ISLR) and then went on to Ng’s course. It was when I viewed the latter that it really clicked. I still think ISLR was good though. Ng uses Matlab (bad if you’re getting used to R), provides better explanations for the math (good), has quizzes (good) and focuses more on modern approaches including neural networks (good). So I would suggest at least enrolling in the Coursera course in order to have it available as a backup.
Re math prerequisites, Khan Academy is also how I got to being able to understand machine learning. I also did a Coursera course in Calculus, which might help. This Jim Fowler is a good teacher, similarly to Sal Khan. To start learning machine learning, you’ll need basic linear algebra—how to multiply vectors and matrices basically. For the introductory stuff, you can apply it without understanding calculus, although it’s fundamental to how machine learning works and you’ll need to learn some calculus sooner or later. But if it helps with motivation, if you know the basics of how to multiply vectors and matrices, you can definitely start Ng’s course, and learn some of the rest in tandem. In the long-run, to get good at ML clearly requires proficiency with linear algebra, calculus and statistics, as well as a general comfort with math that seems to come from practice, so it seems like we have to put in a bunch of solid hours into getting these foundations. More hours per week seems good.
I think that the several hundred dollars it costs you to make a flight there, and a week of lost salary would pay itself off in expected future earnings from the option of working there later, new professional contacts, new insights into cause prioritisation, fun, new friends, extra passion for doing good, etc. The new Global EA Summit will be coming up too, which might be a reason to get down there. You meeting people there just feels like all-round good news to me.
Makes sense. I think I’ll try that in the same order as well.
I don’t actually know the basics of multiplying vectors and matricies (I learned them in college but forgot soon afterward), so I should learn that first.
You’ve convinced me of two changes to make:
First, I should go in sequence with my learning rather than parallel. I think I’ll aim for Khan Academy Algebra → Khan Academy Calculus I → Khan Academy Calculus II → Khan Academy Linear Algebra → Introduction to Statistical Learning → Angrew Ng’s course. (I think I’ll still do Advanced R --> Learn Hadoop in parallel, though, because my R skills are somewhat unrelated to my MR skills.) (To-do for self: re-arrange learning list.)
Second, I should spend more than 2hrs/wk on this. I can probably cut out more EA time. (To-do for self: think on this more.)
-
Yeah, I’ll come out for the Global EA Summit and you’ve convinced me to try to make a full week of it. We have a pretty flexible vacation policy here, so I shouldn’t even lose salary. I just have to reconcile this with other vacation I plan on taking. (To-do for self: plan out vacation for 2015, watch for SF EA Summit dates.)
I’m thinking of also going to SF when Joey and Xio get around to visiting SF, but I don’t know if that’s going to be in 2015.
Excellent,
Good luck!