Winter School AI4A: Artificial Intelligence for Archaeologists, with Python

The Winter School “Artificial Intelligence for Archaeologists, with Python” (AI4A) will be held both online and in person (can be joined online or in person, equivalently) from February 2 to February 13, 2026, organized by the Department of Civilisations and Forms of Knowledge of the University of Pisa, Italy.

The Summer School will enable participants to explore and visualize data with Python, set up and train neural networks from scratch and/or by modifying pre-trained networks (transfer learning), in order to perform classification tasks based on images and/or tabular data. It is built around a new paradigm, which takes into consideration archaeologists as both producers and users of digital archaeological data. One of the school activities will be dedicated to effective prompting strategies, encouraging a critical and informed use of Large Language Models, such as ChatGPT, Gemini or Copilot. AI4A lasts 60 hours.

Archaeology deals with the study of the human past, conducted through material remains, i.e. artefacts that were manufactured, used, and discarded in ancient times. One of the most important task is to classify the artefacts, determining chronology, cultural attribution, form, function and other features. Neural networks and deep learning are powerful tools for supporting and facilitating such tasks, often time consuming and heavily depending on prior knowledge and expertise.

The AI4A school illustrates the use of neural networks for analyzing and classifying multimodal data, such as images, tables, texts. It is conducted, with an hands-on approach, through Python, one of the main programming languages of AI and Data Science, including a wide variety of deep learning tools and network architectures. In order to effectively conduct and support analysis and classification of data coming from tables, images and texts, modern archaeologists should be able to deal with concepts and tools related to new technologies. Such skills are not present in a standard archaeology background, though they are fundamental even to effectively interact with ICT experts.

MAPPA Lab also manages ArcheologicaData, an open-access scientific journal that publishes original contributions as datasets accompanied by articles. Participants, if interested, have the opportunity to publish datasets and related analyses free of charge for both authors and readers. More information on https://www.mappalab.eu/en/archeologica-data/

Target Attendees

Students, graduates, PhD candidates, and post-docs in archaeology or related to Cultural Heritage. The course is open to EU (including University of Pisa students) and non-EU applicants. For an effective learning environment, the number of participants will be limited to 40.

ECTS: 6
Fees: 500 Euros (accommodation and food not included)
Further Info: please email (both)
Professor Gattiglia: gabriele.gattiglia@unipi.it (Scientific Manager)
Dr. Dubbini: nevio.dubbini@unipi.it (Operations Manager)

Deadline for application: January 2, 2026

Training Modules

Why Python

What is Python
Python in data science
Python ecosystem
IDEs and APIs for python

Technical intro

Python and Colab
Python functions
Python libraries
Datasets
Python variable types
Data import and export
Python programming
OpenCV and image pre-processing
Effective prompting strategies for LLMs (ChatGPT, Gemini, Copilot)

Data visualization

Good principles of visualizations
Python packages for data visualization
Scatter plots, histograms, boxplots and complex visualization

Neural networks

Theory on neural networks, forward and back propagation, loss functions, activation functions, gradient descent
Pytorch
Feedforward neural networks
Convolutional Neural Networks
Transfer Learning
Application to classification of images

Timetable:

from Monday to Friday, from 9 am to 1 pm and from 2 pm to 4 pm (CET)

Nevio Dubbini

Data scientist, Ph.D. in Applied Mathematics, having always worked in strongly interdisciplinary contexts, both in business and in academia. He has gained a long-standing experience in mathematical and statistical modelling, machine learning and data analysis software, applied to a variety of sectors, spanning from humanities to social and life sciences, and going through healthcare, IT and educational. He specializes in mathematical models interpreting archaeological data, predictive modelling, and archaeological potential. He has authored about 20 papers appeared in peer reviewed journals and conference proceedings, edited a volume, and has delivered about 20 talks in international conferences. More about Dr. Dubbini on his linkedin profile.

Gabriele Gattiglia

He obtained his Specialisation in Archaeology in 2003 and his PhD in 2010. He is a Researcher in Archaeological Method and Theory at the University of Pisa. He teaches Sources, tools, and methods for Archeology and Digital Archaeology. He leads the MAPPA Lab, which manages the MOD, the Italian repository for Open Archaeological Data. He is the coordinator of ArchAIDE (3-years European H2020 research project), aimed at creating a new system for the automatic recognition of archaeological pottery. He is one of the leading Italian expert in open archaeological data, and GIS, RDBMS and predictive models specialist. Dr. Gattiglia has been the coordinator of MAPPA project (funded by Regione Toscana, Italy), having created a predictive model of the archaeological potential of an urban area. He conducted as director 12 archaeological excavations and 4 archaeological surveys, participated in 100+ archaeological excavations. He published 11 books and 70 papers on national and international peer review journals or conference proceedings. More about Dr. Gattiglia on his twitter and on his academia.edu profiles.

Quirino Saraceni

Quirino Saraceni is a Ph.D. candidate in Artificial Intelligence at the University of Pisa with a research project aimed at integrating ArchAIDE (Archaeological Automatic Interpretation and Documentation of cEramics), tool which exploits the capabilities of artificial intelligence for the recognition and classification of archaeological ceramic fragments, with a robotic arm capable of selecting and sorting the fragments; he has also collaborated on the RePAIR project, which focuses on reconstructing Pompeian frescoes through the use of a robotic arm. He holds degrees in Ancient Studies and Aerospace Engineering, both from the University of Pisa. His interests lie at the intersection of humanities and technology, with a focus on applying artificial intelligence and robotics to archaeological research. He has delivered talks in international conferences on the application of computational methods in archaeology. Github profile

How to apply

Follow this link, with all the instructions from the University of Pisa web page: how to apply

How to reach us

By Plane
The “Galileo Galilei” International Airport, in Pisa, is well connected with many Italian and European cities. It is served by both international airlines and low cost carriers. Byt the way, Pisa airport is one of the closest to the city centre in the world: central railway station is about 15 minutes walking!

By Car
Pisa is reachable following the A12/E80 “Genova – Livorno” Highway, or the A11/E6 “Firenze – Mare” Highway. There are two main parking areas where you can leave your car and get a bus to the city centre. If you arrive from north, leave your car in Pietrasantina Parking. If you arrive from East, leave your car in Brennero Parking.

By Train
The main train station of Pisa, “Pisa Centrale”, connects the town to Italian and European destinations through the nodes of Florence, Turin-Genova and Rome.

Useful links
Pisa Buses www.cpt.pisa.it/ (in italian)
International Airport Galileo Galilei www.pisa-airport.com/
Railway Station www.trenitalia.com/, www.italotreno.com/

Bibliography
– Stevens E., Antoga L., Viehmann T., Deep Learning with Pytorch, Manning (2020)