Near- and mid-infrared (NIR/MIR) spectroscopy, crucial for rapid, noninvasive chemical analysis in industrial applications, is increasingly leveraged by machine learning techniques to extract meaningful information from complex spectral data. Nevertheless, the constraints imposed by limited sample sizes and the occurrence of domain shifts present substantial obstacles to the effectiveness of both deep learning and transfer learning methodologies. In this work, we introduce a unified, task-oriented benchmark that consolidates nine open-access spectral datasets, providing a consistent preprocessing pipeline and clearly defined few-shot tasks. The benchmark is segmented into multiple tests exhibiting varying degrees of heterogeneity, thereby facilitating the evaluation of diverse methodologies across a range of contexts characterized by varying levels of complexity. To the best of our knowledge, meta-learning has not previously been applied in the context of spectrometric analysis. In this study, we confirm that it offers superior performance to traditional transfer learning approaches. A systematic comparison of transfer learning and meta-learning techniques reveals that meta-learning consistently outperforms transfer learning—achieving lower errors in both heterogeneous and homogeneous tasks, consistent with our latent-space analysis showing a more structured cross-task embedding geometry that helps explain these gains. Furthermore, we empirically demonstrate that this benchmark can serve as a starting point for pre-training models on new industrial problems, accelerating adaptation and improving final performance using meta-learning. We complement our analysis with established, model-agnostic interpretability tools, enhancing transparency and industrial relevance in regulated domains.

ABSTRACT Age‐at‐death estimation is an arduous task in human identification based on characteristics such as appearance, morphology or ossification patterns in skeletal remains. This process is performed manually, although in recent years there have been several studies that attempt to automate it. One of the most recent approaches involves considering interpretable machine learning methods, obtaining simple and easily understandable models. The ultimate goal is not to fully automate the task but to obtain an accurate model supporting the forensic anthropologists in the age‐at‐death estimation process. We propose a semi‐automatic method for age‐at‐death estimation based on nine pubic symphysis traits identified from Todd's pioneering method. Genetic programming is used to learn simple mathematical expressions following a symbolic regression process, also developing feature selection. Our method follows a component‐scoring approach where the values of the different traits are evaluated by the expert and aggregated by the corresponding mathematical expression to directly estimate the numeric age‐at‐death value. Oversampling methods are considered to deal with the strongly imbalanced nature of the problem. State‐of‐the‐art performance is achieved thanks to an interpretable model structure that allows us to both validate existing knowledge and extract some new insights in the discipline.

The combination of convolutional and recurrent neural networks is a promising framework. This arrangement allows the extraction of high-quality spatio-temporal features together with their temporal dependencies. This fact is key for time series prediction problems such as forecasting, classification or anomaly detection, amongst others. In this paper, the TSFEDL library is introduced. It compiles 22 state-of-the-art methods for both time series feature extraction and prediction, employing convolutional and recurrent deep neural networks for its use in several data mining tasks. The library is built upon a set of Tensorflow + Keras and PyTorch modules under the AGPLv3 license. The performance validation of the architectures included in this proposal confirms the usefulness of this Python package.

Age estimation is a fundamental task in forensic anthropology for both the living and the dead. The procedure consists of analyzing properties such as appearance, ossification patterns, and morphology in different skeletonized remains. The pubic symphysis is extensively used to assess adults’ age-at-death due to its reliability. Nevertheless, most methods currently used for skeleton-based age estimation are carried out manually, even though their automation has the potential to lead to a considerable improvement in terms of economic resources, effectiveness, and execution time. In particular, explainable machine learning emerges as a promising means of addressing this challenge by engaging forensic experts to refine and audit the extracted knowledge and discover unknown patterns hidden in the complex and uncertain available data. In this contribution we address the automation of the decision making process of Todd’s pioneering age assessment method to assist the forensic practitioner in its application. To do so, we make use of the pubic bone data base available at the Physical Anthropology lab of the University of Granada. The machine learning task is significantly complex as it becomes an imbalanced ordinal classification problem with a small sample size and a high dimension. We tackle it with the combination of an ordinal classification method and oversampling techniques through an extensive experimental setup. Two forensic anthropologists refine and validate the derived rule base according to their own expertise and the knowledge available in the area. The resulting automatic system, finally composed of 34 interpretable rules, outperforms the state-of-the-art accuracy. In addition, and more importantly, it allows the forensic experts to uncover novel and interesting insights about how Todd’s method works, in particular, and the guidelines to estimate age-at-death from pubic symphysis characteristics, generally.

The study of the physical properties of porous media is crucial for petrophysics laboratories. Even though micro computed tomography (CT) could be useful, the appropriate evaluation of flow properties would involve the acquisition of a large number of representative images. That is often unfeasible. Stochastic reconstruction methods aim to generate novel, realistic rock images from a small sample, thus avoiding a large acquisition process. In this contribution, we improve a previous method for 3D-to-3D reconstruction of the structure of porous media by applying generative adversarial neural networks (GANs). We compare several measures of pore morphology between simulated and acquired images. Experiments include Beadpack, Berea sandstone, and Ketton limestone images. Results show that our GANs-based method can reconstruct three-dimensional images of porous media at different scales that are representative of the morphology of the original images. Furthermore, the generation of multiple images is much faster than classical image reconstruction methods.

An appropriate visualization of multiobjective nondominated solutions is a valuable asset for decision making. Although there are methods for visualizing the solutions in the design space, they do not provide any information about their relationship. In this paper, we propose a novel methodology that allows the visualization of the nondominated solutions in the design space and their relationships by means of a network. The nodes represent the solutions in the objective space while the edges show the relationships among the solutions in the design space. Our proposal (called moGrams) thus provides a joint visualization of both objective and design spaces. It aims at helping the decision maker to get more understanding of the problem so that (s)he can choose the most appropriate and flexible final solution. moGrams can be applied to any multicriteria problem in which the solutions are related by a similarity metric. Besides, the decision maker interaction is facilitated by modifying the network based on the current preferences to obtain a clearer view. An exhaustive experimental study is performed using four multiobjective problems with a variable number of objectives to show both usefulness and versatility of moGrams. The results exhibit interesting characteristics of our methodology for visualizing and analyzing solutions of multiobjective problems.

Craniofacial superimposition aims to identify a missing person by comparing its skull with photos of possible candidates. Among the difficult tasks involved, this requires superimposing the skull over each photo, matching the pose of the skull with that of the face, a problem known as skull-face overlay (SFO). Several computerized methods for SFO have been proposed, in an effort to relieve forensic experts of this complex, time-consuming, and subjective task. A system to generate artificial SFO data from computed tomography images has been also introduced, providing researchers with data to test and compare their techniques more reliably. This paper improves the state of the art on both fronts. We introduce a novel SFO algorithm that is substantially more accurate, more reliable, and much faster than the existing methods. An extensive experimental study and statistical analysis validates our findings. Moreover, we propose an improved method to simulate SFO data which, by replacing real photos with simulated ones, is able to generate a wider range of scenarios. This module provides complete control over the pose of the subject and the camera parameters, and even the ability to reproduce inter-expert errors in processing the input data, leading to a more controlled and thorough testing under realistic conditions.

In this paper, we present an evolutionary trust game to investigate the formation of trust in the so-called sharing economy from a population perspective. To the best of our knowledge, this is the first attempt to model trust in the sharing economy using the evolutionary game theory framework. Our sharing economy trust model consists of four types of players: a trustworthy provider, an untrustworthy provider, a trustworthy consumer, and an untrustworthy consumer. Through systematic simulation experiments, five different scenarios with varying proportions and types of providers and consumers were considered. Our results show that each type of players influences the existence and survival of other types of players, and untrustworthy players do not necessarily dominate the population even when the temptation to defect (i.e., to be untrustworthy) is high. Our findings may have important implications for understanding the emergence of trust in the context of sharing economy transactions.

The study of human remains suffers from a lack of information for determining a reliable estimation of the age of an individual. One of the most extended methods for this task was proposed in the twenties of the past century and is based on the analysis of the pubic bone. The method describes some age changes occurring in the pubic bone and establishes ten different age ranges with a description of the morphological aspect of the bone in each one of them. These descriptions are sometimes vague and there is not a systematic way for using the method. In this contribution we propose two different preliminary fuzzy rule-based classification system designs for age estimation from the pubic bone that consider the main morphological characteristics of the bone as independent and linguistic variables. So, we have identified the problem variables and we have defined the corresponding linguistic labels making use of forensic expert knowledge, that is also considered to design a decision support fuzzy system. A brief collection of pubic bones labeled by forensic anthropologists has been used for learning the second fuzzy rule-based classification system by means of a fuzzy decision tree. The experiments developed report a best performance of the latter approach.

Image registration (IR) or image alignment is a fundamental step in medical image analysis when multiple images are involved. In most of such applications, the registration is performed following the intensity-based approach, which turns IR into a complex, computationally expensive, continuous optimization problem. In this paper, we introduce a new technique for intensity-based medical IR using the Bacterial Foraging Optimization Algorithm (BFOA), a novel bio-inspired metaheuristic. BFOA has recently obtained promising results in many real-world applications, including feature-based IR. The new algorithm is compared on a complex medical IR application against recent, outstanding IR techniques both traditional and based on meta-heuristics. The results show that our proposal is competitive with the state of the art, making BFOA a promising solution to tackle other complex, real-world optimization problems.

Craniofacial superimposition involves the process of overlaying a skull with a number of ante-mortem images of an individual and the analysis of their morphological correspondence. Within the craniofacial superimposition process, the skull-face overlay stage focuses on achieving the best possible overlay of the skull and a single ante-mortem image of a missing person. This technique has been commonly applied following a nonautomatic trial-and-error approach. Automatic skull-face overlay methods have been developed obtaining promising results. In this paper, we present two new variants that are an extension of existing 3-D-2-D methods to automatically superimpose a skull 3-D model on a facial photograph. We have modeled the imprecision related to the facial soft tissue depth between corresponding pairs of cranial and facial landmarks which typically guide the automatic approaches. As an illustration of the model's performance, the soft tissue distances associated to studies for Mediterranean population have been considered for dealing with this landmark matching uncertainty. Hence, we directly incorporate the corresponding landmark spatial relationships within the automatic skull-face overlay procedure. We have tested the performance of our proposal on 18 skull-face overlay instances from a ground truth data set obtaining valuable results. The current proposal is thus the first automatic skull-face overlay method evaluated in a reliable and unbiased way.

This paper describes a hybrid level set approach to medical image segmentation. The method combines region-and edge-based information with the prior shape knowledge introduced using deformable registration. A parameter tuning mechanism, based on Genetic Algorithms, provides the ability to automatically adapt the level set to different segmentation tasks. Provided with a set of examples, the GA learns the correct weights for each image feature used in the segmentation. The algorithm has been tested over four different medical datasets across three image modalities. Our approach has shown significantly more accurate results in comparison with six state-of-the-art segmentation methods. The contributions of both the image registration and the parameter learning steps to the overall performance of the method have also been analyzed.

Image registration is a fundamental step in combining information from multiple images in medical imaging, computer vision and image processing. In this paper, we configure a recent evolutionary algorithm for medical image registration, r-GA, with an offline automatic parameter tuning technique. In addition, we demonstrate the use of automatic tuning to compare different registration algorithms, since it allows to consider results that are not affected by the ability and efforts invested by the designers in configuring the different algorithms, a crucial task that strongly impacts their performance. Our experimental study is carried out on a large dataset of brain MRI, on which we compare the performance of r-GA with four classic IR techniques. Our results show that all algorithms benefit from the automatic tuning process and indicate that r-GA performs significantly better than the competitors.

Image registration (IR) is a challenging topic in both the computer vision and pattern recognition fields; its main aim is to find the optimal transformation to provide the best overlay or fitting between two or more images. Usually, the success of well-known algorithms, such as iterative closest point, highly depends on several assumptions, e.g., the user should provide an initial near-optimal pose of the images to be registered. In the last decade, a new family of registration algorithms based on evolutionary principles has been contributed in order to overcome the latter drawbacks. However, their performance highly depends on carefully tuning (usually by hand) the control parameters of the algorithm, which is an error-prone and a time-consuming task. In this paper, we propose a new self-adaptive evolution model to deal with IR problems. To our knowledge, this is the first time a self-adaptive approach has been used for tuning the control parameters of evolutionary algorithms tackling computer vision tasks. Specifically, we introduce a novel design of the proposed self-adaptive approach facing pair-wise range IR problem instances, which is a challenging real-world optimization problem. In addition, several classical approaches, as well as state-of-the-art evolutionary IR methods, have been considered for numerical comparison.

Polygonal surface models are typically used in three-dimensional (3D) visualizations and simulations. They are obtained by laser scanners, computer vision systems or medical imaging devices to model highly detailed object surfaces. Surface mesh simplification aims to reduce the number of faces used in a 3D model while keeping the overall shape, boundaries, and volume. In this work, we propose to deal with the mesh simplification problem from an evolutionary multi-objective viewpoint. The quality of a solution is defined by two conflicting objectives: the accuracy and the simplicity of the model. The Non-dominated Sorting Genetic Algorithm II (NSGA-II) is adapted to tackle the problem. We compare the NSGA-II performance with a classical approach and a single-objective implementation. The comparison has been carried out using different datasets.

We introduce a new intensity-based image registration (IR) technique based on a modern, real-coded genetic algorithm. Our proposal is tested on 16 registration scenarios involving real-world MRI medical images. A novel methodological framework to compare heterogeneous IR algorithms is also described. Following such methodology, our algorithm is compared with four well-known IR techniques of different natures. The proposed method is able to improve the results of these techniques in the majority of the scenarios.

The time and space assembly line balancing problem (TSALBP) considers realistic multiobjective versions of the classical assembly line balancing involving the joint optimization of conflicting criteria such as the cycle time, the number of stations, and/or the area of these stations. This industrial problem is very difficult to solve and of crucial importance in the manufacturing context. As TSALBP-1/3 contains a set of hard constraints like precedences or cycle time limits for each station it has been mainly tackled using multiobjective constructive metaheuristics (e.g. ant colony optimization). Global search algorithms in general -and multiobjective genetic algorithms in particular have shown to be ineffective to solve this family of problems up to now. The goal of this contribution is to present a new multiobjective genetic algorithm design, taking the well known NSGA-II algorithm as a base and new coding scheme and specific operators, to properly tackle with the TSALBP. An experimental study on six different problem instances is used to compare the proposal with the state-of-the-art methods.

Craniofacial superimposition (CS) is a forensic process where photographs or video shots of a missing person are compared with the skull that is found. By projecting both photographs on top of each other (or, even better, matching a scanned 3-D skull model against the face photo/video shot), the forensic anthropologist can try to establish whether it is the same person. The whole process is influenced by inherent uncertainty, mainly because two objects of different nature (a skull and a face) are involved. In this paper, we extend our previous evolutionary-algorithm-based method to automatically superimpose the 3-D skull model and the 2-D face photo with the aim to overcome the limitations that are associated with the different sources of uncertainty, which are present in the problem. Two different approaches to handle the imprecision will be proposed: weighted and fuzzy-set-theory-based landmarks. The performance of the new proposal is analyzed, considering five skull–face overlay problem instances that correspond to three real-world cases solved by the Physical Anthropology Laboratory, University of Granada, Granada, Spain. The experimental study that is developed shows how the fuzzy-set-based approach clearly outperforms the previous crisp solution. Finally, the proposed method is validated by the comparison of its outcomes with respect to those manually achieved by the forensic experts in nine skull–face overlay problem instances.

Image registration is a very active research area in computer vision. Image registration methods, aim to find a transformation between two images taken under different conditions. Point matching is an image registration approach based on searching for the right pairing of points between the two images. From this matching, the registration transformation can be inferred by means of numerical methods. In this paper, we tackle the medical image registration problem adapting a new advanced hybrid metaheuristic composed by the GRASP and the evolutionary path relinking algorithms, called G&EvPR. The experiments conducted in this work have shown the good performance of G&EvPR compared to similar approaches of the state of the art when dealing with different medical image modalities. In particular, a good tradeoff between search space diversification and intensification is achieved.

Scatter search is a population-based method that has recently been shown to yield promising outcomes for solving combinatorial and nonlinear optimization problems. Based on formulations originally proposed in the 1960s for combining decision rules and problem constraints, such as the surrogate constraint method, scatter search uses strategies for combining solution vectors that have proved effective in a variety of problem settings. We present a scatter-search implementation designed to find high-quality solutions for the 3D image-registration problem, which has many practical applications. This problem arises in computer vision applications when finding a correspondence or transformation between two computer images obtained under different conditions. Our implementation goes beyond a simple exercise on applying scatter search, by incorporating innovative mechanisms to combine and improve solutions and to create a balance between intensification and diversification in the reference set. Furthermore, heuristic information taken from a preprocessing of the images is incorporated into the algorithm to improve its performance. Our computational experimentation tackling two different medical registration applications establishes the effectiveness of scatter search in relation to different approaches usually applied to solving the problem. We have considered both simulated magnetic resonance images and real-world computerized tomography images as data sets. To measure the robustness of our proposal, the image data sets are intentionally selected for addressing registration environments with the presence of noise, anatomical lesions, and occlusions between images.

Craniofacial superimposition is a process that aims to identify a person by overlaying a photograph and a model of the skull. This process is usually carried out manually by forensic anthropologists; thus being very time consuming and presenting several difficulties in finding a good fit between the 3D model of the skull and the 2D photo of the face. In this paper we present a fast and automatic procedure to tackle the superimposition problem. The proposed method is based on real-coded genetic algorithms. Synthetic data are used to validate the method. Results on a real case from our Physical Anthropology lab of the University of Granada are also presented.

Different tasks in forensic anthropology require the use of three-dimensional models of forensic objects. Since range scanners do not allow to capture the whole object in a single image, multiple scans from different views are needed to supply the information to construct the 3D model. Range image registration methods study the accurate integration of the different views acquired by these scanners. Specifically, pair-wise image registration methods manage every adjacent pair of scanned views. Our proposal is based on the adaptation of a previous work in order to apply the scatter search evolutionary algorithm to pair-wise image registration in forensic anthropology applications. To measure the performance of this adaptation, we design an experimental setup considering some of the most recent and accurate evolutionary techniques for the problem to compose a 3D model of one skull from our Physical Anthropology Lab.

Image registration is a crucial task in many areas such as computer vision, neurosurgery, remote sensing, and cartography. Different approaches to solve the problem have been proposed in the specialized literature but present some drawbacks. In this contribution, we focus our interest on the 3D surface IR problem and our proposal considers the use of an emergent evolutionary global optimization strategy called scatter search, providing a fast and accurate algorithm to estimate similarity transformations. To measure its performance, we make a broad comparison with some of the most accepted and accurate classical and evolutionary techniques. The new proposal will be validated using two different surfaces, a synthetic and a magnetic resonance image, considered by other researchers in the field, both in noise-free and noisy scenarios.