ok, here's the simplified version of batch learning and Online learning. in batch learning.. the system learn the complete set of data at once, also it requires a lot of computing resources and the system is trained and deployed. this is also called offline learning.

what if we need to train new data??
yep.. if we have to feed new data, we have to completely retain the system with the new data from the scratch. the new system then replaces the older version. this is a time consuming and with the increase in dataset the required resource can be way costly and resource intensive.

and when it comes to Online learning, the system may learn from new data's incrementally as we feed the system. and then, it also works well with a limited computation. the learning rate determines how quickly the system learns from the data.. but it can quickly forgot the older information.

how ever, while training in online data.. feeding biased data can decrease the performance over time. so, its necessary to monitor the flow of the data and systems performance.

and, finally choosing batch learning and online learning depends upon the application that we choose to work on!

- I plan to write simplified breakdowns of technical concepts related to AI and robotics. Make sure to follow me!