FedCurv-DR Enables Federated Continual Learning for Cultural Heritage

Ioannis Theologitis, Debin Meng, Stylianos Eleftheriadis, Vasileios Lolis, Konstantinos Votis· August 21, 2026 View original

Key takeaways

  • Federated Continual Learning is well-suited for distributed and evolving cultural heritage data.
  • FedCurv-DR is a lightweight FCL strategy that reduces forgetting.
  • It balances performance, fairness, and energy efficiency for sustainable AI.
  • The method protects learned knowledge while minimizing communication overhead.

Who benefits

Cultural HeritageDigital HumanitiesEducationAI EthicsData Privacy

Summary

FedCurv-DR is a lightweight, regularisation-based federated continual learning strategy designed for cultural heritage data, which is often distributed and evolving. It reduces forgetting and balances performance, fairness, and energy efficiency by accumulating parameter-importance estimates across clients and experiences.

Artificial intelligence holds significant potential for cultural heritage and digital humanities, particularly for large-scale retrieval and analysis of digitized collections. However, data in this sector is typically distributed across various institutions, subject to ownership and access restrictions, and constantly evolving. Federated Continual Learning (FCL) is an ideal paradigm for this context, allowing models to learn from distributed and sequential data without requiring raw collections to be shared. This paper introduces FedCurv-DR, a lightweight FCL strategy based on regularization. FedCurv-DR works by accumulating estimates of parameter importance across different clients and learning experiences, which helps protect previously acquired knowledge. These estimates are updated only at fixed intervals to minimize communication and computational overhead. Evaluated on the WikiArt image dataset for genre classification with evolving styles, FedCurv-DR demonstrates reduced forgetting and a balanced approach to performance, fairness, and energy efficiency, promoting sustainable AI in cultural heritage.

Why it matters

For professionals in cultural institutions, digital humanities, and AI ethics, FedCurv-DR offers a practical and sustainable way to leverage AI for distributed, sensitive, and evolving datasets while respecting data privacy and resource constraints.

How to implement this in your domain

  1. 1Explore federated learning solutions for collaborative AI projects involving sensitive or distributed data.
  2. 2Assess FedCurv-DR's potential for applications requiring continual learning without data centralization.
  3. 3Implement strategies to measure and balance performance, fairness, and energy efficiency in AI models.
  4. 4Collaborate with cultural heritage institutions to pilot federated learning approaches for digital archiving and analysis.

Original post by Ioannis Theologitis, Debin Meng, Stylianos Eleftheriadis, Vasileios Lolis, Konstantinos Votis

"arXiv:2608.20038v1 Announce Type: new Abstract: Artificial intelligence can support cultural heritage and digital humanities through large-scale retrieval and analysis of digitized collections. However, cultural heritage data are often distributed across institutions, constrained…"

View on X

Originally posted by Ioannis Theologitis, Debin Meng, Stylianos Eleftheriadis, Vasileios Lolis, Konstantinos Votis on X · view source

Want to go deeper?

Turn these trends into skills with Learnijoy's hands-on AI & tech courses.

Explore courses

More in AI Research

AI ResearchAI Engineering & DevTools

Decoding Silent Reading from Non-Invasive EEG

This research demonstrates that open-vocabulary word-level and semantic information can be reliably decoded from non-invasive EEG during silent reading. Using a contrastive decoder and a large dataset from a single participant, the study shows decoding scales log-linearly with training data and extends to rare words.

Ingo Marquardt, Anthilia Alchanat, Priyanka JainAug 21, 2026
AI ResearchAI Engineering & DevTools

Exact Learning Coefficients for Singular Models

This paper presents the first deterministic algorithm for exactly computing local learning coefficients (Real Log Canonical Thresholds) for two-dimensional singular models. This breakthrough provides ground truth for calibrating sampling-based estimators and reveals algebraic structure in learning coefficients, outperforming sampling in shallow regimes.

Gr\'egoire Sergeant-Perthuis (CQSB, Sorbonne Universit\'e), Elias Tsigaridas (Ouragan Team, INRIA), Jules Tsukahara (Ouragan Team, INRIA)Aug 21, 2026
AI Engineering & DevToolsAI Research

Standardized ML Evaluation for Power System Protection

This paper proposes a standardized framework for evaluating machine learning applications in power system protection, addressing inconsistencies in current research. It defines seven critical study dimensions and instantiates the framework with a case study on fault classification and localization using a public benchmark.

Julian Oelhaf, Georg Kordowich, Paula Andrea P\'erez-Toro, Christian Bergler, Johann J\"ager, Andreas Maier, Siming BayerAug 21, 2026