Title:OCTOPUS: Overcoming Performance andPrivatization Bottlenecks in Distributed Learning

Abstract:The diversity and quantity of the data warehousing, gathering data from distributed devices such as mobile phones, can enhance machine learning algorithms' success and robustness. Federated learning enables distributed participants to collaboratively learn a commonly-shared model while holding data locally. However, it is also faced with expensive communication and limitations due to the heterogeneity of distributed data sources and lack of access to global data. In this paper, we investigate a practical distributed learning scenario where multiple downstream tasks (e.g., classifiers) could be learned from dynamically-updated and non-iid distributed data sources, efficiently and providing local privatization. We introduce a new distributed learning scheme to address communication overhead via latent compression, leveraging global data while providing local privatization of local data without additional cost due to encryption or perturbation. This scheme divides the learning into (1) informative feature encoding, extracting and transmitting the latent space compressed representation features of local data at each node to address communication overhead; (2) downstream tasks centralized at the server using the encoded codes gathered from each node to address computing and storage overhead. Besides, a disentanglement strategy is applied to address the privatization of sensitive components of local data. Extensive experiments are conducted on image and speech datasets. The results demonstrate that downstream tasks on the compact latent representations can achieve comparable accuracy to centralized learning with the privatization of local data.