Key Challenges of Machine Learning Model Deployment

One of the main challenges of deploying your model into production is, concept and data drift. Loosely, this means what if your data changes after your system has already been deployed? Let's take two examples of this before defining them specifically to have a better intuition of how this might look in real life. For the first example, assume that you are working at a mobile manufacturing company and you have trained a learning algorithm to detect scratches on smartphones under one set of lighting conditions, and then maybe the lighting in the factory changes. Let's walk through a second example using a speech recognition task.