Privacy Aggregation of Teacher Ensembles - Object detection
Does anyone know how PATE(Privacy Aggregation of Teacher Emsembles) can be applied to object detection? This becomes quite complex since we're dealing with possibly multiple objects detected in the same image and we need to aggregate those multiple labels predicted by the teacher models. Could someone help me understand how PATE can be applied in object detection?
Topic privacy computer-vision deep-learning machine-learning
Category Data Science