1
$\begingroup$

I came across Incremental Learning algorithms paper, where incremental algorithms are compared. I have problem with general understanding. Will the model be updated /adapts itself automatically when new data comes in?

Does it know by itself that new data has arrived and it learns?

In general, can anyone explain how training, testing, and model adaption is carried out with such incremental algorithms?

Improve this question
$\endgroup$
2
  • 1
    $\begingroup$ your link doesn't work (looks like a path to a local file) $\endgroup$
    – oW_
    Commented Jun 7, 2019 at 15:18
  • $\begingroup$ @oW_ i have edited $\endgroup$
    – priya
    Commented Jun 12, 2019 at 7:58

1 Answer 1

Reset to default
1
$\begingroup$

Incremental learning is analogous to online learning. It is based on the assumption that your model can receive a continuous stream of data from which it can keep learning indefinitely. Training is therefore based on Mini-Batch Gradient Descent optimization: you feed batches of data into the Network as soon an new data comes in.

Hope this helps, otherwise let me know.

Improve this answer
$\endgroup$
3
  • 1
    $\begingroup$ not all online learning algorithms are based on mini-batch gradient descent, e.g. there are incremental versions of decision trees $\endgroup$
    – oW_
    Commented Jun 7, 2019 at 15:21
  • $\begingroup$ Mini-Batch gradient descent is mandatory for online Deep Learning. Géron's book Hands-On Machine Learning With Scikit-Learn and Tensorflow, in my opinion the best book on ML ever written, said this too. Also, what you called incremental versions of decision trees must be based on feeding mini-batches of data into the model. Ok, certain model don't use gradient descent, but feeding mini-batches is mandatory for online / incremental learning. $\endgroup$
    – Leevo
    Commented Jun 7, 2019 at 15:28
  • $\begingroup$ Would the performance be the same as just retraining the whole model from scratch with the new data? Is this concept mainly used for speed purposes? $\endgroup$
    – Isbister
    Commented Nov 9, 2019 at 10:35

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.