Studying artificial intelligence?

Question

Machine learning

Hello. After reading many answers and articles, I could not find a really worthwhile answer with my introductory.
Most of the answers to questions about artificial intelligence focus on the need for the questioner to start learning mathematics. The fact is that at the moment I am getting a fairly strong mathematical education at the Faculty of Software Engineering and I want to start moving in the field of artificial intelligence. The industries that interest me most are the development of AI in games, as well as neural networks. In this regard, I want to ask a question about how you started studying this topic? What tasks to set for yourself at the beginning, what technologies to choose for this, and everything like that.
ps Now I am actively studying C and it would be a big plus if this knowledge can be applied at the beginning of my studies

Answer 1

strong enough mathematical education

Lol: D
If I turned a lot of answers, then I think I saw: https://cache-mskstoredata04.cdn.yandex.net/downlo...
I xs, maybe it's a super university like ITMO or something else, but I'm not on one The IT department has not seen a similar mathematics program.
Given the fact that you want to do artificial intelligence - add a couple more HEX to that list, because this is too little explored area.
In machine learning, matstat, theorver, and linal are enough.
And if you want to study this:
That first one is not even AI, there are ordinary search algorithms, etc., for example, the A-star algorithm works wonders if you use the loss probability metric as a heuristic.
The second is just a graph, where linear adders are defined on the vertices. Plus a bit of swearing like gradients and stuff. Studied under articles on Habré. Now there are even more of them.
I do not advise, too many details to which it will be necessary to pay attention. Memory, going beyond the limits of the array, and generally unknown garbage with the processing of tensors (multidimensional arrays, often 3-dimensional and 4-dimensional, sometimes five-dimensional) - you yourself understand that you can break your leg in C.
I advise python, an elementary language, it is studied in a couple of days with proper knowledge of languages ​​​​like si and pluses, because intuitively everything will be simple, that you ask the question: "WHAT COULD BE THIS? YYY"

Answer 2

It all depends on your goals. If you need a quick start, then neither C nor knowledge of mathematics is needed at all, even at the level of the fifth grade. Everything has long been out of the box in the form of libraries for R and Python. You feed the algorithm the necessary data and get the result. There is no point in doing it all yourself from scratch, well, only if you do not pretend to develop some new machine learning algorithm.

Answer 3

The structure of neural networks - Matlab documentation.
Recognition:
a) in the same place - Computer Vision Toolbox, highlighting features (features)
b) statistics: analysis, classification, clustering (there are also examples in Matlab)

Answer 4

As the course on the courser from Andrew En already advised, plus the shad lectures for 2014 on machine learning are on the torrents. The fact is that

sufficiently strong mathematical education at the faculty

and what is needed to USE neural networks and machine learning may not coincide very much. Let me give you a little advice, for the program of networks and learning algorithms, use a python for some kind of java. And also try to study why all this crap is needed and what can be done at the level of using libraries. If the desire persists, then start digging deeper into networking and mathematical features and see how it works inside.
At the expense of AI in games, I don’t remember on which of the specialized forums I saw the phrase, but it very much reflects what is happening in the gaming industry right now

at the moment, AI is used very rarely in games, since good scripts do an excellent job with their tasks, and people on the other side of the monitor cannot distinguish AI from a script