Original article was published on Deep Learning on Medium
What is Batch and Online Learning?
[ML0to100] — S1E7
Another criterion used to classify Machine Learning systems is whether or not the system can learn incrementally from a stream of incoming data.
1. Offline (Batch) Learning — You start with a certain amount of data. That might be a sample or it might be all of your available data. Regardless, that is all the data you’ll use to train your model. Once you have a model you’re satisfied with, you start making predictions. Over time you’ll determine that your model’s performance has degraded to a point where you feel it is time to re-train your model using more/newer data and features.
2. Online learning — You’re continuously updating your models with new batches of data. You’re not re-training the model as you would in offline learning, you’re simply updating the model weights based on new observations.
Both of these systems are discussed in detail in the consecutive episodes.
Read Next- What is Offline/Batch Learning? [S1E8]