as long as I know, online learning takes actions at each time step (for each data), and one-pass algorithm just can see each data once. I already read Wikipedia: about streaming algorithms.

These algorithms have many similarities with online algorithms since they both require decisions to be made before all data are available, but they are not identical. Data stream algorithms only have limited memory available but they may be able to defer action until a group of points arrive, while online algorithms are required to take action as soon as each point arrives.

I still fail to understand how they are fundamentally different. both of them just see each data ones? and the difference is just about memory?

and I want to know is there any popular one-pass algorithm exist which I can compare my result with it? cause I already compare my result with online algorithms and I don't know if there is any difference or not?


1 Answer 1


From what I understood, like you said, one pass learning is basically learning by seeing the data once. So if I learn by taking data as a single instance, mini-batch or large batches as long as I go over them once (epoch=1) they qualify as one pass learning.

With regards to online learning I simply like to think of it as learning and predicting in real time. So for an input stream, if I update the model at every time step it is online learning. If for some reason I again input the same stream it is still online learning but not one pass learning.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.