Self-Supervised Learning -- Part 1

Over the past decade, supervised machine learning has solved some of the most challenging real-world problems. Supervised machine learning has impacted complex real-world challenges ranging from object detection in scenes, speech recognition, machine translation, human pose detection, medical image segmentation and numerous other high-impact real-world problems. Fundamentally, the deep learning process consists of several stages; where in the first stage, we transform the real work problem into a machine learning problem. Subsequently, depending on the task, the next step is to collect a rich and diverse set of labelled examples for the problem. E.g. for the issue of detecting objects in a scene, the system will need labelled data from real-world areas specifying the bounding boxes around the things of interest.