قابل توجه نویسندگان محترم، مقالاتی که از تاریخ 1404/07/13 برای نشریه ارسال می شوند، شامل پرداخت هزینه بررسی نخواهند شد.
Semi-Annual
Back to the articles list |
Back to browse issues page
1- University of Tehran , ma.kabirian@gmail.com
2- University of Tehran
3- Noor Genetics Research Center
2- University of Tehran
3- Noor Genetics Research Center
Abstract: (11 Views)
Sex estimation of human skeletal remains is one of the fundamental steps in bioarchaeology and forensic anthropology, providing crucial demographic information about past populations and assisting in forensic identification cases. Traditional methods, such as the Buikstra-Ubelaker standard, offer valuable frameworks but suffer from subjective interpretations, interobserver variability, and regional morphological differences. To address these challenges, this study not only develops two fuzzy inference systems (FIS) designed to estimate the sex of skeletal remains based on morphological features, but also aims to highlight the potential of fuzzy logic in anthropological and archaeological studies.
The first fuzzy system is constructed according to the Buikstra-Ubelaker standard, treating selected cranial traits equally and using predefined feature weights. The second system is developed based on empirical osteological data collected from 128 Iranian skeletal samples, integrating both cranial and pelvic features. Fuzzy rules were extracted systematically using decision tree modeling, enhancing the data-driven nature of the second system.
Both fuzzy inference systems were developed using a Mamdani-type architecture and implemented in the Python programming environment. Validation of the models involved applying the first system to a dataset of 50 individuals and the second system to a dataset of 19 individuals, all with sex estimations previously made by experts and confirmed through DNA analysis. The Buikstra-Ubelaker-based system achieved an accuracy rate of 48%, whereas the empirical-data-based system demonstrated 83% accuracy. Findings indicate that fuzzy inference systems, particularly those built on empirical, population-specific data, can provide more accurate and objective sex estimations than standard-based methods. Further refinement in membership function design, feature weighting, and dataset expansion is recommended to optimize performance and enhance the applicability of fuzzy systems in archaeological and forensic contexts.
The first fuzzy system is constructed according to the Buikstra-Ubelaker standard, treating selected cranial traits equally and using predefined feature weights. The second system is developed based on empirical osteological data collected from 128 Iranian skeletal samples, integrating both cranial and pelvic features. Fuzzy rules were extracted systematically using decision tree modeling, enhancing the data-driven nature of the second system.
Both fuzzy inference systems were developed using a Mamdani-type architecture and implemented in the Python programming environment. Validation of the models involved applying the first system to a dataset of 50 individuals and the second system to a dataset of 19 individuals, all with sex estimations previously made by experts and confirmed through DNA analysis. The Buikstra-Ubelaker-based system achieved an accuracy rate of 48%, whereas the empirical-data-based system demonstrated 83% accuracy. Findings indicate that fuzzy inference systems, particularly those built on empirical, population-specific data, can provide more accurate and objective sex estimations than standard-based methods. Further refinement in membership function design, feature weighting, and dataset expansion is recommended to optimize performance and enhance the applicability of fuzzy systems in archaeological and forensic contexts.
Keywords: Fuzzy Inference System, Sex Estimation, Archaeological Anthropology, Forensic, Osteological Data Analysis
Technical Note: Original Research |
Subject:
Archaeometry
Received: 2025/04/27 | Accepted: 2025/09/23
Received: 2025/04/27 | Accepted: 2025/09/23
| Rights and permissions | |
 |
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License. |
