Optimizing and Managing Machine Learning Models for Edge Deployment

Every Machine Learning (ML) practitioner knows that the ML development life cycle is an extremely iterative process, from gathering, exploring, and engineering the right features for our algorithm, to training, tuning, and optimizing the ML model for deployment. As ML practitioners, we spend up to 80% of our time getting the right data for training the ML model, with the last 20% actually training and tuning the ML model. By the end of the process, we are all probably so relieved that we finally have an optimized ML model that we often don’t pay enough attention to exactly how the resultant model is deployed. It is, therefore, important to realize that where and how the trained model gets deployed has a significant impact on the overall ML use case. For example, let’s say that our ML use case was specific to Autonomous Vehicles (AVs), specifically a Computer Vision (CV) model that was trained to detect...