Artificial intelligence provides greater accuracy in the classification of modern and ancient bone surface modifications

Manuel Domínguez-Rodrigo,Enrique Baquedano,Marcos Pizarro-Monzo,Natalia Abellán,Blanca Jiménez-García,Elia Organista,Gabriel Cifuentes-Alcobendas

doi:10.1038/s41598-020-75994-7

Abstract

Bone surface modifications are foundational to the correct identification of hominin butchery traces in the archaeological record. Until present, no analytical technique existed that could provide objectivity, high accuracy, and an estimate of probability in the identification of multiple structurally-similar and dissimilar marks. Here, we present a major methodological breakthrough that incorporates these three elements using Artificial Intelligence (AI) through computer vision techniques, based on convolutional neural networks. This method, when applied to controlled experimental marks on bones, yielded the highest rate documented to date of accurate classification (92%) of cut, tooth and trampling marks. After testing this method experimentally, it was applied to published images of some important traces purportedly indicating a very ancient hominin presence in Africa, America and Europe. The preliminary results are supportive of interpretations of ancient butchery in some places, but not in others, and suggest that new analyses of these controversial marks should be done following the protocol described here to confirm or disprove these archaeological interpretations.

Highlights

Bone surface modifications are foundational to the correct identification of hominin butchery traces in the archaeological record
In the past ten years, analytical tools for analyzing bone surface modifications (BSM) have become very sophisticated, involving the use of 2D and 3D geometric morphometric analyses[8,9,10,11,12,13,14], uni- and bivariate metric analyses through 3D reconstruction of BSM15,16, multivariate qualitative analysis using frequentist and bayesian traditional techniques[17,18,19,20], machine learning analyses[21,22], machine learning techniques combined with geometric m orphometrics[23] and, most recently, artificial-intelligence computer vision through deep learning (DL)[24]
Each of the seven DL models tested provided a high accuracy in the classification of the three types of BSM of the testing set (Table 1)

Summary

Introduction

Bone surface modifications are foundational to the correct identification of hominin butchery traces in the archaeological record. We present a major methodological breakthrough that incorporates these three elements using Artificial Intelligence (AI) through computer vision techniques, based on convolutional neural networks This method, when applied to controlled experimental marks on bones, yielded the highest rate documented to date of accurate classification (92%) of cut, tooth and trampling marks. The DL method that we present here is the first one and it has yielded higher accuracy rates than any previous taphonomic approach in differentiating structurally similar and dissimilar BSM jointly This provides a compelling referential base with which controversial cut marks in the archaeological record can be more solidly identified and interpreted. The application of this referential image database to some selected important BSM in the archaeological record provides some preliminary interpretations that challenge current interpretations on the earliest presence of humans in some major geographical areas and the earliest traces of butchery in the archaeological record

Methods

Results

Conclusion