Decoding the Past: An Uncertainty-Aware Deep Learning Framework for Sex Attribution in Prehistoric Hand Stencils

Karel Becerra; Boris Mederos; Dean Snow; Ramón A. Mollineda

cs.CVAug 14, 2026

Decoding the Past: An Uncertainty-Aware Deep Learning Framework for Sex Attribution in Prehistoric Hand Stencils

Authors: Karel Becerra, Boris Mederos, Dean Snow, Ramón A. Mollineda

Organizations: Data Science and AI Division, Azyri, Miami, Florida, USA · Departamento de Física y Matemáticas, Instituto de Ingeniería y Tecnología, Universidad Autónoma de Ciudad Juárez, Ciudad Juárez, Mexico · Professor Emeritus of Anthropology at Pennsylvania State University, Pennsylvania, USA · Institute of New Imaging Technologies, Universitat Jaume I, Castelló de la Plana, Spain

Abstract

Determining the biological sex of the individuals who created Upper Paleolithic hand stencils remains a challenging problem due to the absence of ground truth, population differences between contemporary and prehistoric groups, and the uncertainty introduced by image degradation. Traditional morphometric methods suffer from high structural overlap across sexes, poor cross-population generalizability, and subjective feature engineering. This study presents an uncertainty-aware deep learning framework for sex attribution in prehistoric hand stencils that explicitly models, propagates, and aggregates uncertainty throughout the analytical pipeline. The methodology combines dual image processing, dual contour extraction, structured silhouette augmentation, model architectural diversity, and ensemble-based decision aggregation. The pipeline generates twelve plausible silhouette realizations per stencil to capture boundary uncertainties, which are processed by two ensembles of ten deep neural networks each (EfficientNet-B3 and MobileViT-S) trained on 14,036 contemporary hand samples. Furthermore, a triangulated validation scheme integrates ensemble predictions with unsupervised 2D latent-space manifold mapping (UMAP + k-NN) and explainable AI spatial attributions (LayerCAM) to ensure anatomical consistency. On contemporary data, ensemble models achieve strong classification performance, with accuracies exceeding 88% in older age groups. When applied to prehistoric stencils, the framework produces both sex predictions and confidence measures of internal agreement, enabling the distinction between morphologically stable and ambiguous cases. Convergence across ensemble predictions, latent-space structure, and interpretability analyses shows that uncertainty can become a measurable component of archaeological inference, enabling robust and reproducible decoding of ancient rock art.

Figures & tables

Panel of Figure 1 — Figure 1: Four prehistoric hand images from El Castillo cave ( Snow, 2006 ) .

Reference	Training data	Test data	Features extraction	Decision rule	Protocol
( Fernández-Navarro et al., 2025b )	Contemporary	Paleolithic	Automatically learned	DNN	Cross-domain
( Fernández-Navarro et al., 2025a )	Contemporary	Simulated, Paleolithic	Automatically handcrafted	LDA	Cross-domain
( Mollineda et al., 2025 )	Contemporary	Paleolithic	Automatically learned	DNN	Cross-domain
( Rabazo-Rodríguez et al., 2017 )	Contemporary	Paleolithic	Manually handcrafted	LDA	Cross-domain
( Nelson et al., 2017 )	Contemporary	Contemporary	Automatically handcrafted	Fisher’s LDA	Leave-one-out
( Permana et al., 2015 ) .	Contemporary	Paleolithic	Manually handcrafted	Nearest class mean	Cross-domain

Table 1: Studies on sex attribution in prehistoric hand stencils, categorized by the type of hand data used (simulated stencils, Paleolithic stencils, or contemporary hand images). Abbreviations: LDA = Linear Discriminant Analysis; DNN = Deep Neural Network; SVM = Support Vector Machine.

Figure 2: Uncertainty-aware inference system: quantifying sexual dimorphism in prehistoric cave hand stencils. Generated using Gemini Notebook from authors’ prompts and methodology.

	females	males	total
training set	5,778	6,833	12,611
validation set	652	773	1,425
test set	100	100	200

Table 2: Summary of RSNA Bone Age Challenge dataset.

Figure 3: Representative samples of contemporary test X-ray images (top row) and their corresponding extracted silhouettes (bottom row). Each column displays a unique case identified by its 4-digit ID above, with associated sex and age metadata provided below.

Panel of Figure 4 — Figure 4: Bone age distribution by sex across the training, validation, and test subsets of the RSNA Bone Age Challenge dataset. Source: ( Mollineda et al., 2025 ) . Reproduced with permission.

Panel of Figure 5 — Figure 5: Five prehistoric hand images analyzed in the exploratory study ( Mollineda et al., 2025 ) . Labels identify the source caves.

Figure 6: Silhouette extraction and augmentation pipeline (phase 1). From left to right: manual contouring is performed on both original and enhanced hand stencils, followed by the extraction of two primary silhouettes and their subsequent geometric normalization. The data augmentation phase comprises four binary operations (AND, OR, AVG, and morphological mid-set) between the extracted silhouettes. Finally, 3D composition operators combine the six resulting variants (two primary + four binary augmentations) into three-channel representations. Generated using Gemini Notebook from authors’ prompts and methodology.

Panel of Figure 7 — Figure 7: Manual contouring applied to original versus enhanced cave hand stencil images.

Figure 8: Interpretable 2D manifold generated by one of the ten EfficientNet-B3 models, using only the three silhouette variants $S_{and}$ , $S_{or}$ , $S_{mid}$ per stencil to reduce visual clutter.

Architecture	Model	Training	Validation	Test data
Architecture	Model	data	data	All ages	[0–6) y	[6–12) y	[12–19] y
EfficientNet-B3	Mean $\pm$ SEM	$82.6\pm 0.7$	$75.8\pm 0.3$	$81.7\pm 0.4$	$68.1\pm 1.9$	$81.5\pm 1.0$	$85.3\pm 0.6$
	Sum aggregation	$85.4$	$77.7$	$\mathbf{85.7}$	$\mathbf{76.2}$	$\mathbf{85.4}$	$\mathbf{88.4}$
	Max aggregation	$\mathbf{85.9}$	$\mathbf{78.1}$	$84.7$	$\mathbf{76.2}$	$84.3$	$87.2$
MobileViT-S	Mean $\pm$ SEM	$80.3\pm 1.4$	$74.8\pm 0.7$	$80.6\pm 1.7$	$66.7\pm 4.1$	$79.4\pm 1.9$	$85.2\pm 1.6$
	Sum aggregation	$84.2$	$77.0$	$82.7$	$66.7$	$82.0$	$87.2$

Table 3: Performance of EfficientNet-B3 and MobileViT-S on contemporary data, featuring age-stratified results for the test set and overall results for training and validation sets. Results are reported as mean accuracy

\pm

Standard Error of the Mean (SEM) for individual model instances, alongside ensemble accuracy achieved via sum and max aggregation rules. Square brackets indicate inclusion of the boundary values, whereas parentheses indicate their exclusion.

Figure 9: Confusion matrices summarizing class-wise correct and incorrect predictions across age-stratified test data for both ensembles (architectures) under the sum-aggregation rule.

	$k$ -NN		EfficientNet-B3
Cave Image	Predicted Class	SSR +	Predicted Class	SSR/ACM ∗
img_26	Female	0.982	Female	1/0.61
img_28	Female	0.966	Female	1/0.67
img_31	Male	0.697	Male	0.87/0.18
img_38	Male	0.821	Male	1/0.29
Cosquer	Female	0.989	Female	1/0.73

Table 5: Aggregated

k

-NN classification performance within 2D manifolds across the 12 silhouette variants and the 10 EfficientNet-B3 instances.

Figure 10: Mean SSR (support for the predicted class across all silhouette variants) as a function of $k$ . A sensitivity analysis was conducted over the range $k\in[1,50)$ . Since SSR remains consistently above 0.5, the $k$ -NN predictions are shown to be robust and largely insensitive to the choice of $k$ .

Figure 11: Robust LayerCAM attribution maps obtained by geometric median aggregation across 12 silhouette variants and 10 EfficientNet model instances.

Cultural Heritage

Cultural heritage uses AI to document, analyze, preserve, and interpret artifacts, sites, archives, and living traditions.

See all 57 papers

Probabilistic Dating of Historical Manuscripts via Evidential Deep Regression on Visual Script Features
May 7, 2026Ranjith ChodavarapuHistorical Document Analysis Evidential Deep Learning
Kent State University
Biological Sex Determination in Cadavers Using Deep Learning Algorithms from Computed Tomography Images of Pelvis and Skull
Jun 21, 2026Giovanna Herculano Tormena, Davi Nascimento Araújo, Germano Coimbra Soares de Carvalho +10Medical Image Classification
São Carlos School of Engineering, University of São Paulo, São Carlos 13566-590, Brazil · 212Aristoclides Teixeira Institute of Forensic (IMLAT), Goiânia 74.425-030, Brazil · 1São Carlos School of Engineering, University of São Paulo, São Carlos 13566-590, Brazil
TextileNet: Towards Zero-shot Text-style Segmentation of Manuscripts
Jul 10, 2026Anguelos Nicolaou, Antonella Ambrosio, Desiree Di Donato +1Visual Representation Learning Authorship Attribution
University of Graz, Graz, Austria · Università degli Studi di Napoli Federico II, Naples, Italy

May 7, 2026J/K move · Enter open · S save

Probabilistic Dating of Historical Manuscripts via Evidential Deep Regression on Visual Script Features

Ranjith Chodavarapu

Kent State University

Historical Document Analysis Evidential Deep Learning Cultural Heritage Uncertainty Calibration Document Image Understanding Historical Document Processing

Open paper AskPDF arXiv

We introduce a probabilistic approach for dating historical manuscript pages from visual features alone. Instead of aggregating centuries into classes as is standard in the previous literature, we pose dating as an evidential deep regression problem over a continuous year axis, allowing our neural network to output a full predictive distribution with decomposed aleatoric and epistemic uncertainty in a single forward pass. Our architecture combines an EfficientNet-B2 backbone with a Normal-Inverse-Gamma (NIG) output head trained with a joint negative-log-likelihood and evidence-regularization objective. On the DIVA-HisDB benchmark (150 pages, 3 medieval codices, 151,936 patches), our model scores a test MAE of 5.4 years, well below the 50-year century-label supervision granularity, with 93% of patches within 5 years and 97% within 10 years. Our approach achieves PICP=92.6%, the best calibration among all compared methods, in a single forward pass, outperforming MC Dropout (PICP=88.2%, 50 passes) and Deep Ensembles (PICP=79.7%, 5 models) at

5\times

lower inference cost. Uncertainty decomposition shows aleatoric uncertainty is a strong predictor of dating error (Spearman

ρ=0.729

), and a selective prediction about the most certain 20% of patches can provide 0.5 years MAE. We show that predicted uncertainty increases as image degradation worsens, spatial decomposition maps explain which script regions cause aleatoric uncertainty, and page-level aggregation reduces MAE to 4.5 years with

ρ=0.905

between uncertainty and page-level error.

Abstract

Cultural Heritage

Explore similar work

Probabilistic Dating of Historical Manuscripts via Evidential Deep Regression on Visual Script Features

Biological Sex Determination in Cadavers Using Deep Learning Algorithms from Computed Tomography Images of Pelvis and Skull

TextileNet: Towards Zero-shot Text-style Segmentation of Manuscripts