Multi-task learning on the edge: cost-efficiency and theoretical optimality
This work addresses cost-efficiency for edge computing applications, but it appears incremental as it builds on existing multi-task learning and SPCA methods.
The paper tackled the problem of distributed multi-task learning on edge devices by proposing a supervised principal component analysis-based algorithm, achieving significant energy gains with no performance loss in experiments on synthetic and real benchmark data.
This article proposes a distributed multi-task learning (MTL) algorithm based on supervised principal component analysis (SPCA) which is: (i) theoretically optimal for Gaussian mixtures, (ii) computationally cheap and scalable. Supporting experiments on synthetic and real benchmark data demonstrate that significant energy gains can be obtained with no performance loss.