Self-Supervised Learning in Multi-Task Graphs through Iterative Consensus Shift
This work addresses the need for more efficient multi-task learning without costly labeled data, though it appears incremental as it builds on existing multi-task and self-supervised approaches.
The paper tackles the problem of expensive supervision in multi-task learning by introducing a self-supervised method using pseudo-labels and a consensus shift algorithm, resulting in significant improvements across iterations and outperforming recent methods on two challenging datasets.
The human ability to synchronize the feedback from all their senses inspired recent works in multi-task and multi-modal learning. While these works rely on expensive supervision, our multi-task graph requires only pseudo-labels from expert models. Every graph node represents a task, and each edge learns between tasks transformations. Once initialized, the graph learns self-supervised, based on a novel consensus shift algorithm that intelligently exploits the agreement between graph pathways to generate new pseudo-labels for the next learning cycle. We demonstrate significant improvement from one unsupervised learning iteration to the next, outperforming related recent methods in extensive multi-task learning experiments on two challenging datasets. Our code is available at https://github.com/bit-ml/cshift.