The next evolution of AI focuses on empowering edge devices to bring intelligence to environments likeretail stores, smart cities, and hospitals. Developing AI for edge devices is challenging due to limited compute,memory, and connectivity. The tutorial introduces the Open Edge Platform, a suite of tools for building scalable AIsolutions from edge to cloud. It includes Edge AI Libraries for model training, inference, and end-to-enddevelopment, and Edge AI Suites, which provide ready-to-use solutions, sample code, and reference designs fordomains such as retail, manufacturing, smart cities, and robotics.
The tutorial introduces FLAI, a decentralized federated learning protocol combining trusted executionenvironments (TEEs), blockchain coordination, and sPEG scoring to ensure fairness, auditability, and robustnessin collaborative AI. Attendees will learn how enclaves enable secure aggregation and attestation, smart contractshandle task registration and settlement, and how the system matches FedAvg performance while improvingfairness. A live demo will showcase an eHealth case study using wearable and EHR data for privacy-preservingreal-world evidence. By the end, participants will understand how to design and deploy TEE-backed decentralizedAI networks with cryptographic trust and tokenized incentives for applications in health, IoT, and finance.
Smart consumer electronics have become more widespread, they face increasing cyberattacks—ranging fromminor breaches (e.g., smart doorbells) to serious risks (e.g., medical devices). This tutorial demonstrates how tobuild a machine learning-based intrusion detection system that analyzes network traffic to detect malicious activity.It covers data preprocessing, building a pipeline of classifiers from traditional algorithms (Decision Trees, RandomForest) to deep learning models, and validating results with appropriate metrics. The second half includes hands-on experiments to build and test the system in real time. The session concludes with deployment considerationsand future directions.
This tutorial addresses challenges in AI related to privacy, security, and centralization, highlighting FederatedLearning (FL) as a solution for training models on distributed data without exposing raw information. As FLintroduces issues like verifying model integrity and preventing malicious inputs, the tutorial introduces zk-FOLcombining Zero-Knowledge Proofs (ZKPs) and blockchain to ensure privacy, correctness, and trust throughout thefederated learning lifecycle. Architectural innovations, cryptographic techniques for anonymity and confidentiality,and real-world applications in healthcare, finance, and IoT will be presented with future research directions forprivacy-enhancing technologies.
This tutorial will introduce attendees to the rapidly evolving field of Green Streaming and Sustainable Media, focusing on technologies and strategies for reducing energy consumption and carbon emissions across the digital media supply chain. Participants will learn about the environmental impact of streaming, the importance of sustainability in media production, distribution, and playback, and the latest measurement frameworks and AI-driven optimization techniques. Key topics include: - The energy and carbon footprint of streaming services and devices - Regulatory frameworks (e.g., CSRD, Paris Agreement) and their implications for media companies - Technical solutions for sustainable encoding, distribution, and playback - Measurement approaches for energy consumption and CO2 emissions - AI-powered tools for real-time energy prediction and optimization - Case studies and demos, including the FAMIUM GreenView and Wattlify solutions Attendees will gain actionable insights into implementing energy-efficient workflows, understanding regulatory requirements, and leveraging AI for sustainability in media. The tutorial will feature interactive demos and practical examples, empowering participants to drive sustainability initiatives in their organization.
The convergence of communication, computation, and visual intelligence has driven demand for high-qualityimage and video transmission. While current compression standards (JPEG, HEIF, VVC) are efficient, theystruggle with errors over noisy channels. Quantum communication offers a paradigm shift by leveragingsuperposition and entanglement for robust, high-fidelity multimedia delivery with superior noise tolerance andefficiency. This tutorial explores quantum principles for visual data transmission, including single- and multi-qubitencoding, quantum frequency-domain transformations, and error resilience. Attendees will learn how thesetechniques can revolutionize next-generation consumer electronics—such as UHD TVs, AR/VR headsets, andmobile processors—by improving bandwidth, energy.
CTSoc AdministratorCharlotte Kobert charlotte.kobert@ieee.org
