Communication Efficient Coreset Sampling for Distributed Learning

Distributedly learning through wireless network becomes one of the future features with the growth of computation power for devices. Communication becomes the bottleneck for such distributed framework. In this paper, distributed learning is studied using the approach of coreset. In the context of classification, an algorithm of coreset construction is proposed to reduce the redundancy of data and thus the communication requirement, similarly to source coding in traditional data communications. The coreset based sampling is robust to adversary distribution, thus leading to potential applications in distributed learning systems. Both theoretical and numerical analyses are provided to demonstrate the proposed framework.