Can we use sigmoid activation function for regression task where i have to predict a value between 0 and 1?
sure, that’s called logistic regression
Just make sure that you’re using a Binary Cross-Entropy loss, as the MSE loss can have some issues with the Sigmoid (source)