Backpropagation neural network for sex determination from patella in forensic anthropology

Abstract

Forensic anthropology is a discipline that concerned on postmortem identification from skeletal remains in sex determination. In sex determination, besides empirical techniques such as Discriminant Function Analysis (DFA), Artificial Intelligence techniques such as Artificial Neural Network (ANN) should be considered to get more accurate result. This paper proposes back propagation ANN model for sex determination. By using data and DFA result from previous work, this paper compares the result with the result of ANN model obtained from the experiment. A total sample data of 113 patellae has been generated based on statistics values of previous study. The data is divided into three groups of ages (young, middle, and old) and is measured using three parameters (width, height, and thickness). The ANN model produces average accuracy until 96.1% compared to 92.9% result from DFA technique. This concludes that ANN produces more accurate result in sex determination compared to DFA.