how does one get started with Artificial intelligence?
State what's of your interest. AI is a really wide term.
In general for a layman I would advise to firstly learn programming, then some basics of algorithms, data structures and from now on pursue into AI.
If you can't program in anything, go for Python. If you know shit about algorithms/data structures, you will overall be capable of doing simple stuff.
https://intelligence.org/research-guide/
>>8075851
That recursion.
>>8075797
>State what's of your interest.
not him but ive been very interested in neural networks control systems, but its always been sort of a crapshoot for me when ive used them, just wondering if thers a more formal way of working one than just randomly tweaking it.
>>8075894
this fucking guy stole the image i made
you unoriginal fuck
>>8075911
get off of 4chan
>>8075894
Who cares that it's simple if it works ?
Bet you like overengineered german cars, too.
>>8075851
That's not modern at all, it's a bunch of symbolic bullshit
OP, learn about deep learning (
Theory:
start with perceptron, regularization, backprop, multilayer perceptrons, CNN, RNN, then read recent papers
Practical aspect:
Learn python, do a kaggle with Sklearn, then learn Keras and implement some recent models
>>8076043
The early stuff is still useful
>>8076050
Nope, I studied the "recent" early stuff and it's really shitty, pure symbolic learning should be eradicated
>>8075656
Fuck a computer
hundreds of billions of if statements every millisecond
AIML
Artificial Intelligence Markup Language.
You can be a complete computer noob and learn.
>>8075894
And guess what, it fucking works.
>>8077263
Woah, that's deep, man
>>8075851
Agreed that this book is a meme. Started reading it and it is mostly the most narrow AI. Most of it is not even about self-improving systems.
currently reading deeplearningbook.org. You can thank me later.
>>8075894
It's funny because all you unimaginative cucks who told me "machine learning is about manual feature selection" are now getting replaced by an algorithm.
You are probably also victim to the "singularity is not completely certain -> we should not bother working on it" error, no?
>>8078253
Yup. Just printed it out.
>>8078253
Thanks, I'm going to read this.
>>8075894
The best part of this is always the X/Y axis of the neural networks
>>8075656
Get PhDs in neurology and computer science.
>>8075656
Depends what you want to learn to do. What kind of intelligent things?