EdgeDL 2021: 4th International Workshop on

EdgeDL: Deep Learning on Edge for Smart Health and Wellbeing Applications

Irvine, Orange County, California, USA, 23 August, 2021 (Virtual Workshop)

Web: https://sites.google.com/view/ieee-edgedl-2021/

EDAS Paper Submission Link: https://edas.info/N28334

Wiki CFP: TBD

EdgeDL 2021 will happen virtually to eliminate travel requirements. All accepted and presented EdgeDL 2021 papers will be published in IEEE Xplore Digital Library

Co-located with the IEEE International Conference on Smart Computing (SMARTCOMP 2021)

SMARTCOMP is the premier IEEE conference on smart computing

Wearable Internet of Things (wIoT) together with deep learning is revolutionizing the smart health and wellbeing applications. Predominantly IoT devices are good at acquiring medical data and later sending it cloud. Edge-based deep learning infuse intelligence in terms of processing, analysis and inference on edge near wearables. Edge deep learning not only offload the cloud but also ensure high-throughput, low-latency solutions. With edge deep learning, the data is processed on edge leading to improved privacy and security as now the data is not transferred to cloud for inference. Resource constraints on the edge and endpoint IoT devices pose challenges in adopting deep learning solutions. Systems and algorithms deployed in health and fitness devices require research on efficient approaches for signal sensing, analysis and prediction. Recently, deep learning models are increasingly deployed on wearable and edge devices for neural prediction and inference. Modern smartwatches and smart textiles are health as well as fitness device. Deep learning on edge also allows for personalization of medical solutions that enhances the user’s experience. Increasingly more wearables in health and fitness now rely on voice-based assistants. Recently, several custom chips with medical machine learning functionalities are developed to further advance edge deep learning. We live in exciting times when wearables and deep learning are growing in parallel and together creating tremendous impact on smart health & fitness devices, systems and services.

This workshop invites researchers from academia and industry to submit their current research for fostering academia-industry collaboration. The scope of this workshop includes but not limited to the following topics:

• Resource-constrained deep learning for sensing, analysis, and interpretation in IoT healthcare

• Low latency inference on edge for smart health

• Privacy-preserving machine learning, federated learning, on-device training, and differential privacy for smart health

• Knowledge transfer and model compressions for smart health

• Hardware optimization, sparsity, quantization, power savings and algorithmic trade-offs for on-device training and inference

• Recent advances in Edge, Fog and Mist computing for machine learning in healthcare

• Context-aware pervasive system and E2E deep learning for health and fitness applications

• Scalability, privacy, and security aspects of edge-based machine learning

• Emerging applications of edge intelligence in for personalized health and fitness monitoring, tracking and control

• Self-supervised, semi-supervised and unsupervised ML for smart health

• Edge-coordinated health data analysis, visualization and interoperability

• Role of big data in edge-based machine learning for smart health & fitness applications

• Edge devices with custom hardware for Neuromorphic AI and cognitive computing in smart health