Knowledge Assembly: Semi-Supervised Multi-Task Learning from Multiple Datasets with Disjoint Labels
This addresses the challenge of performing multiple tasks simultaneously in real-world scenarios where labeled data is incomplete, offering a practical solution for applications like surveillance and robotics.
The paper tackles the problem of multi-task learning when datasets have disjoint labels by proposing Knowledge Assembly, a semi-supervised method that leverages unlabeled data with model augmentation for pseudo-supervision, achieving improvements of 4.2% for person re-identification and 0.9% for pedestrian attribute recognition over single-task fully-supervised baselines.
In real-world scenarios we often need to perform multiple tasks simultaneously. Multi-Task Learning (MTL) is an adequate method to do so, but usually requires datasets labeled for all tasks. We propose a method that can leverage datasets labeled for only some of the tasks in the MTL framework. Our work, Knowledge Assembly (KA), learns multiple tasks from disjoint datasets by leveraging the unlabeled data in a semi-supervised manner, using model augmentation for pseudo-supervision. Whilst KA can be implemented on any existing MTL networks, we test our method on jointly learning person re-identification (reID) and pedestrian attribute recognition (PAR). We surpass the single task fully-supervised performance by $4.2\%$ points for reID and $0.9\%$ points for PAR.