Federated autoencoder for explainable childhood communication disorder screening / Ali Rougab

Public ISBD Titre : Federated autoencoder for explainable childhood communication disorder screening Type de document : document multimédia Auteurs : Ali Rougab, Auteur ; Abdelmadjid Ben Arfa, Directeur de thèse Editeur : Laghouat : Université Amar Telidji - Département d'informatique Année de publication : 2026 Importance : 57 p. Accompagnement : 1 disque optique numérique (CD-ROM) Note générale : Option : Artificial intelligence and data science Langues : Anglais (eng) Mots-clés : Federated learning  Explainable artificial intelligence  Autoencoder  anomaly detection  Childhood communication disorders  FLAMENCO dataset  Privacy-preserving machine learning  SHAP  Class imbalance  Pediatric screening Résumé : Communication disorder screening in pediatric populations is important for supporting timely clinical assessment and improving developmental outcomes [1]. However, privacy requirements and data variation across care centers make it difficult to build large centralized datasets for training robust deep learning models [2, 3]. To address these challenges, this thesis proposes a federated deep learning framework for childhood communication disorder risk prediction, enhanced with Explainable Artificial Intelligence (XAI). The proposed system uses federated learning (FL) to train a shared model across distributed institutions without exchanging sensitive patient records [4, 5]. It also integrates explainability methods, particularly SHAP, to provide global and local interpretations of model behavior [6, 7]. The framework formulates risk assessment as reconstruction-error-based anomaly detection, which is suitable for imbalanced screening datasets [8, 9, 10]. Evaluation on the FLAMENCO dataset demonstrates the feasibility of the approach, with the federated model achieving strong performance and improving on the Pavlidis et al. baseline in both centralized and federated settings [11, 9]. Overall, the work combines privacy preservation, anomaly detection, and interpretability to support transparent AI-assisted pediatric screening. note de thèses : Mémoire de master en informatique Federated autoencoder for explainable childhood communication disorder screening [document multimédia] / Ali Rougab, Auteur ; Abdelmadjid Ben Arfa, Directeur de thèse . - Laghouat : Université Amar Telidji - Département d'informatique, 2026 . - 57 p. + 1 disque optique numérique (CD-ROM). Option : Artificial intelligence and data science Langues : Anglais (eng) Mots-clés : Federated learning  Explainable artificial intelligence  Autoencoder  anomaly detection  Childhood communication disorders  FLAMENCO dataset  Privacy-preserving machine learning  SHAP  Class imbalance  Pediatric screening Résumé : Communication disorder screening in pediatric populations is important for supporting timely clinical assessment and improving developmental outcomes [1]. However, privacy requirements and data variation across care centers make it difficult to build large centralized datasets for training robust deep learning models [2, 3]. To address these challenges, this thesis proposes a federated deep learning framework for childhood communication disorder risk prediction, enhanced with Explainable Artificial Intelligence (XAI). The proposed system uses federated learning (FL) to train a shared model across distributed institutions without exchanging sensitive patient records [4, 5]. It also integrates explainability methods, particularly SHAP, to provide global and local interpretations of model behavior [6, 7]. The framework formulates risk assessment as reconstruction-error-based anomaly detection, which is suitable for imbalanced screening datasets [8, 9, 10]. Evaluation on the FLAMENCO dataset demonstrates the feasibility of the approach, with the federated model achieving strong performance and improving on the Pavlidis et al. baseline in both centralized and federated settings [11, 9]. Overall, the work combines privacy preservation, anomaly detection, and interpretability to support transparent AI-assisted pediatric screening. note de thèses : Mémoire de master en informatique

---

1 (1 - 1 / 1) Par page : 25 50 100 200