Εξηγήσιμη Τεχνητή Νοημοσύνη (XAI) : ανάπτυξη ευφυούς ψηφιακού βοηθού για την οινοποίηση ερυθρών οίνων

Αρβανιτίδης, Αθανάσιος

Explainable AI (XAI) for red wine production : development an intelligent digital assistant

Master Thesis

Author

Αρβανιτίδης, Αθανάσιος

Date

2025-08

Abstract

The objective of this work is the investigation and application of Explainable Artificial Intelligence (XAI) methods for predicting red wine quality, particularly as winemaking has reemerged as a sector attracting increasing global investment. Within the research framework, a red wine dataset containing numerous laboratory parameters was analyzed. Beyond developing a reliable and interpretable model to support decision-making during wine production and aging, attention was also given to implementing a web application accessible to the end-user – the winemaker themselves. Within this approach, various artificial neural network (ANN) architectures were implemented, with the Sequential Neural Network proving most optimal, as it achieved the best results with stable predictions. Particular emphasis was placed on result interpretability using the advanced SHAP method (SHapley Additive exPlanations), which enabled understanding of the factors influencing the model's predictions. The analysis of results highlighted specific factors affecting red wine quality, thereby providing valuable information to producers. The findings indicate that integrating XAI techniques can significantly enhance the reliability and acceptance of AI systems in practice, offering not only high accuracy but also the ability to understand the collective reasoning behind each prediction. We hope this research will constitute a step towards developing reliable and transparent decision support systems that can be effectively integrated into daily winemaking practice, improving the quality of the resulting products.

Postgraduate Studies Programme

Προηγμένα Συστήματα Πληροφορικής - Ανάπτυξη Λογισμικού και Τεχνητής Νοημοσύνης

Department

Σχολή Τεχνολογιών Πληροφορικής και Επικοινωνιών. Τμήμα Πληροφορικής

Number of pages

Language

Greek

URI

https://dione.lib.unipi.gr/xmlui/handle/unipi/18208

Collections

Τμήμα Πληροφορικής

Show full item record

Except where otherwise noted, this item's license is described as
Αναφορά Δημιουργού - Μη Εμπορική Χρήση - Παρόμοια Διανομή 3.0 Ελλάδα