Call for Paper
Johannes TINTNER | Bernhard SPANGL | Michael MELCHER
(University of Natural Resources and Life Sciences, Vienna , Austria)
Keywords: tree-based methods, neural networks, support vector machines, self-organizing maps
Call: Machine learning algorithms faced a considerable breakthrough in very different fields of application, where huge amount of data are collected. They started to be developed in the 80s of the last century and can roughly be categorized as either supervised or unsupervised. Representative members of supervised learning are regularization methods for prediction and classification, additive and tree-based models, neural networks and support vector machines. Representative members of unsupervised learning are clustering algorithms, self-organizing maps, principal and independent component analysis and multidimensional scaling.
Modern analytics in archaeometry often produce such huge data sets. Therefore, it seems natural that a rising number of applications can be realized. A literature search on the platform “Scopus” found between 2010 and 2017 each year between one and five publications dealing with machine learning methods in an archaeological context. In 2018 the number increased to 12 and in 2019 to 19. We can assume this trend will proceed exponentially.
Especially spectroscopic methods like multi- or hyperspectral imaging demand advanced statistical methods. But also prediction models for the age of wood have been established with the help of machine learning methods. Classification tasks regarding pattern recognition of pottery, sex determination in human remains or the restoration of ancient texts are typical examples, where methods of machine learning can be applied.The session invites submissions dealing with the application of machine learning methods in archaeometric questions. Please indicate the specific methods applied and why these methods were chosen for the specific problem. Describe the methodological approach in a widely understandable way and discuss advantages and disadvantages in relation to the problem.
Submission (open April 15, 2020)
Mind the guidelines!