BlinkNet: Software-Defined Deep Learning Analytics with Bounded Resources
Abstract
Deep neural networks (DNNs) have recently gained unprecedented success in various domains. In resource-constrained systems such as mobile devices, QoS-aware DNNs are designed to meet latency requirements of mission-critical deep learning applications. However, none of the existing DNNs have been designed to satisfy both latency and memory bounds simultaneously as specified by end-users in the resource-constrained systems. In our research, we propose BLINKNET, a runtime system that is able to guarantee both latency and memory/storage bounds via efficient QoS-aware per-layer approximation. We implement BLINKNET in Apache TVM and evaluate it using Cifar10-quick and VGG16 network models. Our experimental results show that BLINKNET can meet the latency and memory requirements with an average 2% accuracy loss.
About the Presenters
Theresa VanderWeide
Theresa VanderWeide has a bachelor of science degree in Mathematics. She is a first year grad student, pursuing her master’s degree in Computer Science. Her research is focused on resource constrained neural networks.