Network Support for High-performance Distributed Machine Learning
This work addresses the problem of optimizing network topology for distributed machine learning to improve performance and reduce cost for practitioners.
This paper proposes a system model and an algorithm, DoubleClimb, for high-performance distributed machine learning that defines the logical network topology around the learning task. DoubleClimb minimizes learning cost while meeting target prediction error and execution time, achieving a 1+1/|I|-competitive solution and outperforming state-of-the-art alternatives.
The traditional approach to distributed machine learning is to adapt learning algorithms to the network, e.g., reducing updates to curb overhead. Networks based on intelligent edge, instead, make it possible to follow the opposite approach, i.e., to define the logical network topology em around the learning task to perform, so as to meet the desired learning performance. In this paper, we propose a system model that captures such aspects in the context of supervised machine learning, accounting for both learning nodes (that perform computations) and information nodes (that provide data). We then formulate the problem of selecting (i) which learning and information nodes should cooperate to complete the learning task, and (ii) the number of iterations to perform, in order to minimize the learning cost while meeting the target prediction error and execution time. After proving important properties of the above problem, we devise an algorithm, named DoubleClimb, that can find a 1+1/|I|-competitive solution (with I being the set of information nodes), with cubic worst-case complexity. Our performance evaluation, leveraging a real-world network topology and considering both classification and regression tasks, also shows that DoubleClimb closely matches the optimum, outperforming state-of-the-art alternatives.