Research Seminar by Wei Yang | Enabling Efficiency Robustness and Security of Deep Learning Systems at The Edge

Enabling Efficiency Robustness and Security of Deep Learning Systems at The Edge

Deep Neural Networks (DNNs) have shown potential in many applications. However, the power of using DNNs comes at substantial computational costs. The costs, especially the inference-time cost, can be a concern for deploying DNNs on resource-constrained embedded devices such as mobile phones and IoT devices. To enable deploying DNNs on resource-constrained devices, researchers propose a series of deep learning systems where the amount of inference-time computation varies for different inputs. This talk will present challenges and a series of work to deploy energy-efficient and robust deep learning systems on the Edge systems/devices. This talk first reviews some of my past research and then will discuss a few ongoing work towards enabling the wide-scale deployment of resource-constrained embedded AI systems like UAVs, autonomous vehicles, Robotics, IoT-Healthcare / Wearables, Industrial-IoT, etc.