In module-1, I learned linear and logistic regression, in module-2 about neuro network, I realize the neuro network only calculates the sigmoid score, not does any regression, is it right?
You refer to Course 2, but you posted in the CNN topic, which is Course 4.
Can you be more specific?
What is the exact name of the lecture, or the assignment or lab notebook file?
And the question you are asking sounds more like it is about the first two weeks of DLS Course 1. Or maybe the first course in MLS.
Well Logistic Regression also calculates a sigmoid score, since it is doing a binary (“Yes/No”) classification. So the distinction is not between neural networks and other functions, it is about the purpose of the functions in question. If your purpose is to perform a “classification”, meaning that the function identifies the input as belonging to a particular class, then the output of the function will either be a sigmoid value (in the case that it is a binary classification) or a softmax value (in the case that there are more than two possible classes that are being identified).
On the other hand, if the purpose of your function is to predict some kind of number, like a stock price or a temperature, then the function is called a “regression” and the output will typically be a real number in some range that is applicable to the quantity you are predicting. You can use neural networks to predict stock prices as well as to do classifications.
Note that there are some inconsistencies in terminology, e.g. Logistic Regression is a classifier not a regression in the sense that I used in the previous two paragraphs. Sorry, but we just have to deal with some of the historic terminology.