Special Sessions

Special Session on Deployed and Validated Machine Learning Applications in Real-World Environments

This special session aims to provide a forum for presenting and discussing machine learning and data mining applications that have gone beyond model design or offline experimentation and have been actually developed, deployed, used, and/or validated in real-world environments. The session is especially intended for contributions showing how intelligent models have been integrated into operational systems, decision-support tools, recommender systems, digital services, industrial platforms, or organizational workflows, as well as how these systems have evolved after deployment through monitoring, user feedback, retraining, or pipeline-based continuous improvement. 

While scientific venues still receive a majority of submissions focused on methodological novelty, benchmark performance, or offline experimental comparison, and these contributions are essential for scientific progress, there is also a clear need for a complementary space devoted to practice-oriented machine learning research, where the main contribution lies in successful deployment, validation in realistic conditions, user adoption, measurable operational impact, lifecycle management, or lessons learned from practical use. This need is increasingly recognized in leading international venues through applied and industry-oriented tracks that explicitly emphasize deployed systems, post-launch evaluation, and practical constraints. For instance, the KDD2025 Applied Data Science or the ADMA 2024 Industry tracks.

This session is also aligned with IDEAL's scope and with related initiatives already represented in previous IDEAL conferences. Building on these directions, the proposed session places stronger emphasis on contributions where intelligent systems are not only designed for real-world domains, but are also actually deployed, used, validated, or supported by operational pipelines for continuous model improvement as new data become available. Papers are expected to clearly explain the application problem, the target users, the operational context, the deployed or validated solution.

Particularly welcome are submissions that report operational constraints, system integration barriers, maintenance issues, concept drift handling, retraining strategies, data and model governance, or lessons learned from practical implementation. Contributions may additionally discuss how data processing pipelines, real-time or batch inference setups, or feature-serving mechanisms are implemented as part of the deployed solution, provided these are framed within the broader story of real-world operation and validation.

Topics of interest include, but are not limited to:

·       Deployed machine learning and data mining systems in health, business, finance, education, mobility, energy, agriculture, manufacturing, public administration, and smart environments.

·       Intelligent decision-support and recommender systems with evidence of actual use or validation in realistic settings.

·       Lifecycle management of machine learning systems.

·       Pipelines for continuous learning, retraining, performance monitoring and model updating.

·       Engineering aspects of deployed ML systems, including data processing pipelines, batch and real-time inference architectures, and feature-serving components.

·       Real-world case studies reporting impact, engineering trade-offs, and lessons learned.

·       MLOps, data engineering, interfaces, governance, explainability, privacy, and accountability in deployed or deployment-ready systems.

·       Hybrid contributions combining methodological soundness with demonstrated practical operation, validation, or maintainable pipeline design.

Submissions centered only on algorithmic novelty, benchmark comparison, or offline experiments without evidence of deployment, realistic validation, or a credible operational pipeline for continued improvement are outside the main focus of the session.

Organising Committee:

• Antonio Tallón – Universidad de Huelva (UHU)

• Pablo Bermejo – Universidad de Castilla-La Mancha (UCLM)

• Juan Carlos Alfaro - Universidad de Castilla-La Mancha (UCLM)

More special sessions to appear shortly...

© Copyright 2026 IDEAL – All rights reserved