MDLdroid: a ChainSGD-reduce Approach to Mobile Deep Learning for Personal Mobile Sensing

TL;DR

MDLdroid is a decentralized mobile deep learning framework that enables resource-efficient, on-device collaborative learning for personal sensing, significantly improving training speed over traditional methods.

Contribution

The paper introduces MDLdroid, a novel decentralized framework with ChainSGD-reduce and reinforcement learning for efficient on-device deep learning in mobile sensing.

Findings

Model training is 2x to 3.5x faster than single-device training.

Training is 1.5x faster than master-slave approaches.

Effective resource balancing improves training efficiency.

Abstract

Personal mobile sensing is fast permeating our daily lives to enable activity monitoring, healthcare and rehabilitation. Combined with deep learning, these applications have achieved significant success in recent years. Different from conventional cloud-based paradigms, running deep learning on devices offers several advantages including data privacy preservation and low-latency response for both model inference and update. Since data collection is costly in reality, Google's Federated Learning offers not only complete data privacy but also better model robustness based on multiple user data. However, personal mobile sensing applications are mostly user-specific and highly affected by environment. As a result, continuous local changes may seriously affect the performance of a global model generated by Federated Learning. In addition, deploying Federated Learning on a local server, e.g., edge server, may quickly reach the bottleneck due to resource constraint and serious failure by attacks. Towards pushing deep learning on devices, we present MDLdroid, a novel decentralized mobile deep learning framework to enable resource-aware on-device collaborative learning for personal mobile sensing applications. To address resource limitation, we propose a ChainSGD-reduce approach which includes a novel chain-directed Synchronous Stochastic Gradient Descent algorithm to effectively reduce overhead among multiple devices. We also design an agent-based multi-goal reinforcement learning mechanism to balance resources in a fair and efficient manner. Our evaluations show that our model training on off-the-shelf mobile devices achieves 2x to 3.5x faster than single-device training, and 1.5x faster than the master-slave approach.