Why LODAS 2026?
Bridging the gap between model-centric research and system-level deployment realities
LODAS 2026 focuses on AI systems that must operate under real-world constraints such as limited resources, privacy requirements, latency bounds, dynamic environments, and distributed data ownership. The workshop brings together research on learning, optimization, and intelligent computing systems that enable AI to function reliably in edge, federated, multi-agent, and cyber-physical settings.
Recent advances in federated learning, evolutionary computation, human-centered AI, foundation models, agentic AI, and edge/cloud intelligence have highlighted a growing gap between model-centric research and system-level deployment realities. This workshop aims to bridge that gap by emphasizing system-aware AI, where learning algorithms, optimization strategies, and architectural decisions are co-designed to meet operational constraints.
We welcome contributions including empirical studies, benchmarks, simulations, and real-world deployments across domains such as IoT, smart cities, healthcare, robotics, and autonomous systems.
Important Dates
All dates follow the FLICS 2026 main conference schedule. Please refer to the main conference website for the most up-to-date timeline.
Note: All dates are synchronized with FLICS 2026. Please check the main conference website for the latest information.
Topics of Interest
Distributed, Federated, and Edge Learning
- Federated learning algorithms, protocols, and orchestration
- Cross-device vs cross-silo FL; personalization and heterogeneity
- Communication-efficient and resource-aware learning
- Privacy-preserving learning, secure aggregation, DP, robustness
- Trust, resilience, and adversarial threats in decentralized learning
AI Systems, Architectures, and Deployment
- AI system architectures for edge–fog–cloud continuums
- Resource-aware scheduling, placement, and deployment optimization
- Model partitioning, offloading, compression, and distillation
- MLOps for distributed and federated systems
- Performance, scalability, and reliability evaluation of AI systems
Agentic AI and Autonomous Systems
- Multi-agent learning, coordination, and decentralized decision-making
- Agentic workflows, tool-using agents, and autonomous pipelines
- LLM- and VLM-based agents in real systems
- Safety, controllability, monitoring, and evaluation of agentic AI
- Agents for planning, optimization, and adaptive control
Optimization, Scheduling, and System-Level Intelligence
- Multi-objective optimization and Pareto-front analysis for learning systems
- Constrained and safe optimization (hard constraints, feasibility-first, penalty methods)
- Constraint-based AI, CP-SAT, MILP, and hybrid solvers
- Learning-optimization co-design
- Workflow scheduling, DAG optimization, and service placement
- Energy-aware, latency-aware, and cost-aware optimization
Applications and Real-World Settings
- Smart cities and urban computing (mobility, safety, accessibility, energy, sensing)
- Healthcare and public health decision support
- Industry 4.0, robotics, predictive maintenance, and industrial automation
- Education and digital public services
- Environmental monitoring and sustainability analytics
- Public-sector and large-scale infrastructure AI
- Negative results, lessons learned, and real-world failures
Human-Centered and Trustworthy AI
- Fairness-aware learning, bias mitigation, and equity metrics
- Accessibility-by-design and inclusive intelligent services
- Interpretability, transparency, and human-understandable policies/models
- Robustness, calibration, reliability, and uncertainty estimation
- Human-in-the-loop optimization, preference learning, interactive EC/ML
- Auditability and evaluation protocols for real deployments
Foundation Models in Systems Contexts
- Deploying foundation models under system constraints
- On-device / edge adaptation of LLMs and multimodal models
- Retrieval-augmented and hybrid symbolic–neural systems
- Efficiency, scalability, and evaluation of large models in production
EC + ML Methods
- Evolutionary hyperparameter optimization, AutoML, and neural architecture search
- Population-based training, coevolutionary learning, and memetic EC–ML hybrids
- Evolutionary reinforcement learning and policy search
- Surrogate-assisted optimization and simulation-based optimization
Cyber-Physical Systems, Digital Twins, and IoT
- AI for cyber-physical systems and digital twins
- Real-time analytics, monitoring, and anomaly detection
- Simulation-based evaluation, sim-to-real transfer
- IoT data management, streaming, and event-driven intelligence
- Smart infrastructure and industrial systems
Submission Guidelines
Choose the format that best fits your work
Accepted workshop papers will appear in the main conference IEEE proceedings
Format Requirements: All submissions must follow the FLICS 2026 submission guidelines and use the conference template. Please consult the main conference website for detailed formatting instructions. Remember to add keywords to your submission.
Submission Portal: Workshop papers must be submitted through the FLICS 2026 EasyChair system. Make sure to select the appropriate workshop track during submission.
7–8 Pages
Mature research contributions with substantial original work, complete evaluations, and clear contributions to the field.
4–6 Pages
Work-in-progress, preliminary results, vision papers, or position statements on emerging challenges and opportunities.
1–2 Pages
Undergraduate student contributions, early-stage ideas, prototypes, or course/research projects. Great for getting started!
Organizing Committee
For questions about scope, submissions, or participation, please contact any organizer.
Primary Contact: Jamal Toutouh — University of Málaga (Spain)
Use email for questions about scope, submissions, attendance, or volunteering for PC/panels/demos.
Suggested email subject: [LODAS 2026 Workshop] Question
Ready to Bridge the Gap?
Help us bridge the gap between model-centric research and system-level deployment. Your contributions on federated learning, agentic AI, edge systems, and real-world deployments are essential!