Any advice for a beginner?

Community Lounge
1

Hi ! A few months back I became interested in Artificial Intelligence and I ended up reading On Intelligence.That book really has been a big influence.
I have watched all the HTM school episodes till date and also have read an introductory neuroscience textbook(and BAMI).
I wish to get my hands dirty with NuPIC. Unfortunately, my programming knowledge is quite limited as I am not from a computer science background.(I know intermediate level Java ,but that’s about it)
Where should I start?
Also ,What knowledge do I need to understand this theory better?Statistics?Neuroscience?
Please guide me in the right direction :slight_smile:

2

If you’d like, there’s an official Java version of Nupic: https://github.com/numenta/htm.java

If you want to use python, there’s a lot of good python resources out there, though a lot of them focus on python 3 (nupic only works on python2). Here’s a book you can use to learn it: http://greenteapress.com/wp/think-python-2e/

2 Likes
3

htm.java is a great place to start if you want to tinker with HTM using Java, but it is not an “official Java version”. While part of the Numenta Open Source umbrella, htm.java is not maintained or supported by Numenta in any official capacity.

2 Likes
4

Not created by Numenta, but owned by Numenta, and “blessed” by Subutai, which is why it states “Official” on the landing page…

It also went through a 3 month vetting process just recently in which the same Random Number Generator was used in both NuPIC and HTM.Java and got Exactly the same output. So after all that work, and craziness - I’m afraid this point is worth “quibbling” over! :wink:

2 Likes
5

Thanks guys,I’ll start with htm.java.I’ll let you know how it goes!

1 Like
6

If you want to use HTM on the JVM, you use HTM.Java.

1 Like
7

I think neuroscience will just make it more confusing. HTM also doesn’t involve much statistics, at least in how it’s described. The quickest way to learn is probably HTM school. You can also write your own HTM program if you can’t figure out NuPIC. I can’t understand it, but there’s pseudocode available.

8

Yes,I think I’ll find making a simple version of HTM myself much more interesting.It would also help develop my understanding of the theory.(I’m also not very comfortable with Git hahah)

9

I agree. HTM school is awesome.

1 Like