Machine Learning at the Edge

Traditional machine learning pipelines and methods break down in supporting machine learning at the edge; however, the data we get via embedded systems and edge devices is valuable to solve many problems. Rather than taking the traditional approach, we can utilize new distributed data science and machine learning models, such as federated learning, to properly learn from data at the edge. Federated learning can also provide other benefits in de-centralizing our data collection, providing more security for the individual data points and also allowing more individualization of the models on the edge devices.