What is the difference between AI architecture and AI models? Are both of them the same? If not, please distinguish both of them and give example of each.
Well simply put AI model can be seen just as a flowchart showing how the control flow moves where it moves how it moves why it moves etc. However AI architecture refers to the next step after building an AI model AI architecture involves representation of the functions that you use in your program. It also involved declaration of the variables you're going to use, the functions that you are going to use, etc.
Architecture describes a general approach to a ML problem, and the parameterization of that approach. For example, a neural net architecture would define the number and size of different layers, the type of each layer, and so on.
A model is one specific instance of a given architecture, trained on a given dataset. For the example of neural nets, the model would primarily consist of the trained weights of each node from the architecture.
One would use a model to make predictions on new examples at runtime. It is possible to generate many models using the same architecture, even with the same training data; the result of each training process is a new model.
It mostly seems to be a personal preference type of thing. But in my readings, AI architecture typically means a large scale structural difference (connectionist / GOFAI; deep stack / recurrent, while AI models are finer distinctions between methods in a common architecture (say, the AlexNet vs other CNNs)
Although this is a kind of an ambiguous question I am choosing to answer it as I see fit.
First of all, there are 2 parts in creating an AI agent, the logical part and the learning part. The logical part is called AI itself while while the other part is called Machine Learning (in general both ML and AI are considered under AI).
Now ,in AI or any combinatorial designs you need something called a model, a mathematical model specifically. In this model you know exactly how each and every artworks and can easily derive how will it work i its entirety. This type of models are mainly used in the AI part.
In the Machine Learning part, we have architectures. They are not exactly what we call mathematical models. We have scarce understanding of their inner workings. We only know the mathematics behind it. We try to fit 1 architecture to a problem. If it does not solve the problem satisfactorily, we try other architectures.
Specifically, if we create a program for Towers of Hanoi we have to decompose the solution into smaller parts whose mathematical model we have to create. We know the working of each step and the values that will be returned.
Whereas, in a Neural Network architecture, you just create 3-4 layers with some number of nodes, a certain learning rate, a certain cost function, a certain cost reduction technique and hope for the best. You have no realistic way to know the correctness of your thinking
In my opinion, this is the difference between a model and architecture in AI context. In real life also if we see, an architect merely draws the building plan. Its a civil engineer who has to first draw/build up the mathematical model to see how the building weight shall be distributed on the foundations, whether the architecture is practical or not.
As far as AI books are concerned these 2 are the best in their class:
And for Machine Learning you can look into this answer I provided: Vetted sources of AI Theory/Tools/Applications for an experienced programmer new to the field?
As @DuttaA mentioned this is an ambiguous question.
Let me give an example.
We have alphabets in all the language.
To learn any language and interact in that language
We need to
- learn first what the alphabets are in the selected language.
- Then each word and its meaning
- Then what kind of words when used in the what kind of order convey what kind of meaning.
Coming to Artificial Intelligence(Sentence), Deep Learning(Word) and Machine Learning(Alphabet)
We have Machine Learning to extract meaningful info from the data we have.
Deep Learning a concept which has proven itself in the recent times in image recognition,letter recognition and speech. Use Machine Learning concepts.
Artifical Intelligence will definitely have a combination of few of the deep learning concepts.
Hope you can understand after going through all these what the difference between
You can visit Deep Learning for in dept info