Be it DL or classic Machine Learning (ML), when it comes to using models in production, things can get challenging. The main reason is that data fuels ML and data can change over time. When an ML model is deployed in production, it is re-trained at certain intervals as the data keeps changing over time. Therefore, re-training ML is not a luxury but a necessity when you are thinking of production-based purposes. DL is only a sub-field of ML and it is no exception to the previous statements. There are two popular methods that ML models are trained on—batch learning and online learning, especially when they are in production.
We will be discussing online learning in the next section. For this section, let's introduce ourselves to the concept of batch learning. In batch learning, we start by training an ML model on a specific chunk...
