How whould i use this Neural Network Library?

Help and Feedback Scripting Support
#1

So, one day i got an idea how about an ai that tries to learn stuff but in roblox? I researched a bit and i found this post and it apperently made neural networking easier, i tried reading his github page ect but there is no clear explanation on how i whould apply stuff into it… maybe its just me but eh, can someone like explain me how this library works and how i can add algorithms onto it (like geneticAlgorithm)

Thanks so much!!

(Also i do know i already made a post but im switching up the question since i think i can research to get the one i wanted in my last post but rn there is no info on the actual library or atleast one i can find…)

2 Likes
#2

I have read through the documentation, I think that you need to be more specific about what stuff that you want the AI to learn else nobody can answer your question, also the library already has the genetic evolution algorithms.

1 Like
#3

that library just provides math functions, its like me writting a ray tracer but using a vector library to help me with representing the rays. it’s not like aiLibrary.createAi() and it’s useless if you dont even understand what machine learning is

3 Likes
#4

Yeah i agree, it will be useless if i dont know a thing about machine learning… thats why im trying to learn it rn

#5

Well, the only thing im really confused about it how to use the nn library, as @CoderHusk said, it just produces math functions but how whould i use the math functions and how whould i make the ai learn things (like for example, it learned that it has to jump in a certain position how whould i transfer that to the knowledge of the ai) and how whould i set the ai to have objectives (for example, reach a certain part and the closer to the part the better)
Im kinda confused on how this library works, im making this post just to clarify myself…

#6

If I am correct, you do not understand how AI works/learns and in particular what this library is about. Before I start, let me clarify something: Artificial Intelligence is not the same with Machine Learning(ML). Artificial intelligence(AI), is basically the concept of using computers that can imitate human intelligence whereas Machine Learning is a subset of AI covering machines that can learn on their own without needing explicit programming to do so.

Now, there is not one Machine Learning approach, but we generally divide ML approaches into three categories. Neural Networks, are in the category of Supervised Learning (you can search more about it). I will not discuss what Neural Networks are, as there are plenty of articles out there, but I will give you some basic examples so you can have a basic understanding of how this library does work.

Ex1) Some neural networks elements are the following: the Input Layer( a layer that accepts input features and passes on the information to the hidden layer), the Hidden Layer( it performs all sort of computation on the features entered through the input layer and transfer the result to the output layer) and the Output Layer( a layer that brings up the information learned by the network to the outer world).

Ex2) Neural Networks have an activation function that as the name implies, decides whether a neuron should be activated or not by calculating weighted sum and further adding bias with it.

Now, these two examples as you understand, are some key elements of neural networks. What does this mean? That every time you want to use for some reason Neural Networks, you will have to code them. Now, what the Library does, is to have these key elements ready in advance, so you do not spend time re-coding them. Note: The Library has much more features than these, I just want to give some examples to you. In fact, this Library uses and some other algorithms, like the GA.

Also, I see that you are asking how to use these math functions and how the AI(?) would learn. In order to understand this, you need to first understand what a neural network is and how it can be used. You can see this pdf An Introduction to Neural Networks (ed.ac.uk) if you are interested in learning more about this subject.

2 Likes