FI2025Federated Intelligence

Special Session at IEEE SMC 2025 ​

October 5-8, 2025, Vienna, Austria

The convergence of Federated Learning (FL) and Collective Intelligence (CI) represents a new frontier in distributed AI. This special session explores the synergies between these powerful paradigms, seeking innovative approaches to privacy-preserving collaborative intelligence.

Important dates: https://www.ieeesmc2025.org/important-dates.html

Call for Papers: https://www.ieeesmc2025.org/files/content/program/SMC25-SpecialSession-CYB12.pdf

Submission Guidelines: https://www.ieeesmc2025.org/call-for-papers.html

Submission: https://conf.papercept.net/conferences/scripts/start.pl

Session code: 5wvfg

Description ​

The convergence of Federated Learning (FL) and Collective Intelligence (CI) is a very new trend in distributed AI. While FL enables collaborative model training across decentralized devices without direct data sharing, CI leverages the collective wisdom of artificial agents and human experts to solve complex problems. This session aims to explore the exciting potential of integrating these two powerful paradigms. By combining privacy preservation of FL with problem-solving competence of CI, we aim at more robust, adaptive, and responsible intelligent systems leveraging diverse distributed computational architectures and heterogeneous devices for handling large-scale, heterogeneous data in dynamic environments. This enables not only learning from statistically heterogeneous and distributed data, but also leveraging the collective knowledge and decision-making of diverse participants, as well as mitigating the heterogeneity of their technological platforms. This session will delve into the theoretical foundations, practical applications, and emerging challenges of Federated Intelligence (FI). We seek to foster discussions on how these synergistic approaches can address real-world problems in domains such as, but not limited to, healthcare, smart cities, IoT, and social sensing, while respecting privacy and promoting collaboration.

Justification ​

The proliferation of heterogeneous edge devices and the growing demand for privacy-preserving data analysis highlighted the limitations of centralized machine learning. FL emerged as promising solution, but often missing to incorporate diverse perspectives and collaborative problem-solving. CI, on the other hand, provides frameworks for aggregating and leveraging collective knowledge, but typically relies on centralized data in servers and cloud. FL and CI fusion enables FI systems that are both privacy-aware and collectively intelligent, offering key advantages:

• Enhanced Privacy and Security: FL's distributed training safeguards sensitive data, while CI can further enhance privacy by obfuscating individual contributions.
• Improved Model Robustness and Generalization: CI-based knowledge aggregation of diverse agents leads to more robust and generalizable models with enhanced resilience to malicious behavior.
• Scalability and Adaptability: FL's distributed architecture enables scalability to massive datasets and adaptivity to statistically heterogeneous data, while CI meets the dynamic environment and changing user needs, leveraging the heterogeneous expertise of multiple agents.
• Collaborative Problem Solving: CI fosters collaboration of diverse agents, enabling the solution of complex problems that are beyond the capabilities of individual agents.
• Democratized AI: FL empowers edge devices to participate in FI for a more democratized landscape, enhancing trustworthiness, fairness and responsibility.

Topics of Interest ​

• CI-enabled privacy-preserving FL mechanisms (e.g., differential privacy, homomorphic encryption, secure aggregation).
• CI methods (e.g., swarm intelligence, evolutionary computation, ontology alignment) integration in FL.
• Federated meta-learning and transfer learning for personalized and adaptive CI.
• Distributed consensus for FL and CI.
• FI applications in healthcare, smart cities, IoT, social sensing, and other domains.
• Communication-efficient and energy-aware FI in cloud-edge continuum architectures.
• Trust, reputation, and explainability management in FI systems.
• Statistical heterogeneous data handling (e.g. non-IID, structural, correlational, causal, incompleteness, noise) in FI.
• Incentivized participation and collaboration in FI.
• Blockchain-enabled FI for secure and transparent model governance and incentivization.

Organizers ​

Supported By ​

• IEEE SMC Technical Committee on Computational Collective Intelligence
• Romanian Academy of Technical Sciences, Section "Information and Communication Technology – Computers and Telecommunications"
• Department of Computers and Information Technology, Faculty of Automation, Computers and Electronics, University of Craiova, Romania
• Intelligent Distributed Systems Research Group

Contact ​

For inquiries, please contact Costin Bădică at costin.badica@edu.ucv.ro

Web Master: Ionuț Murarețu ionut.muraretu@edu.ucv.ro, University of Craiova, Romania