Welcome to the Edge Intelligence course
Our ever-increasing ability to allocate, process, and extract valuable information from the network's edge triggered new decentralized and broadly distributed data analytic systems. More than 60 billion devices and sensors are expected to be connected to the Internet by 2025 creating massive amounts of data. Alongside the growth of IoT capabilities, Artificial Intelligence (AI) has emerged as the next technology phenomenon. For a long time, centralized deployment models were utilized to facilitate such AI/ML operations. Yet, many modern applications (e.g., autonomous vehicles, smart cities and healthcare, etc.) require making decisions in the right place and at the right time in time! A novel paradigm known as Edge Intelligence comes into play, performing data operations proximity to where data is captured based on artificial intelligence.
Course Objective
This course aims to provide students with an overview of this new paradigm, including edge analytics and intelligence, different types of edge/fog services, distributed data processing, along the latest developments in the field, such as Federated Learning.
This course will cover the foundational aspects of advanced data operation and analytics in edge computing. The course emphasizes unleashing Artificial Intelligence algorithms, models, and applications on the network's edge, aka Edge Intelligence (EI) architectures. Various aspects of distributed models of data-intensive architecture include large-scale distributed analysis over Edge/Fog, and the Internet of Things (IoT) deployment models, will be covered. The course is designed over three logical parts as follow: (I) Foundations of Edge Intelligence, including the associated deployment dependencies of edge networking (e.g., SDN and NFV), communication and network protocols, and the intersection of data and edge QoS. (II) Edge Intelligence operational environment that covers three pillars: data management (e.g., allocation and preparation, privacy), resource management (e.g., allocation, scheduling, and provisioning), and finally, application management (e.g., task placement, load-balancing, offloading, etc.). (III) Edge Intelligence implementations, covering lifecycle (allocation, preparing, caching, processing, and analysis), lightweight machine learning (for edge/IoT nodes), distributed machine learning, and federated learning with privacy-preserving aspects.
IMPORTANT
- For your inquiries, you can reach out to the course Primary Instructor Dr Feras M. Awaysheh at feras.awaysheh@ut.ee.
- Lecturers will be in class unless announced otherwise, but they will be recorded.
- Practices will be fully online and recorded.
- It is not always lecture/practice sequences every week. That is, we might have two lectures in the same week, in some other week we can have two practice sessions in a row. Have a look at the tentative syllabus in the Lectures Section. Any changes will be announced at least one session ahead and on the course Moodle page.
- The course is scheduled as follows:
* Mon. 16.15 - 18.00 weeks 24-39 * Tus. 16.15 - 18.00 weeks 24-39
- Link for lectures and practice sessions: (TBA)!
- Moodle platform (TBA)!
Recommended materials:
- William Stallings, Foundations of Modern Networking, Pearson, 1st edition, Nov. 2015, (ISBN-13: 978-0-13-417539-3; ISBN-10: 0-13-417539-5).
- Xiaofei Wang et. al., Edge AI Convergence of Edge Computing and Artificial Intelligence, Springer 2020, ISBN 978-981-15-6185-6. (eBook) https://doi.org/10.1007/978-981-15-6186-3
- Awaysheh, Feras M., et al. "Big data resource management & networks: Taxonomy, survey, and future directions." IEEE Communications Surveys & Tutorials (2021). https://ieeexplore.ieee.org/abstract/document/9478917