The Capabilities of Boltzmann Machines to Detect and Reconstruct Ising System's Configurations from a Given Temperature.

The restricted Boltzmann machine (RBM) is a generative neural network that can learn in an unsupervised way. This machine has been proven to help understand complex systems, using its ability to generate samples of the system with the same observed distribution. In this work, an Ising system is simulated, creating configurations via Monte Carlo sampling and then using them to train RBMs at different temperatures. Then, 1. the ability of the machine to reconstruct system configurations and 2. its ability to be used as a detector of configurations at specific temperatures are evaluated. The results indicate that the RBM reconstructs configurations following a distribution similar to the original one, but only when the system is in a disordered phase. In an ordered phase, the RBM faces levels of irreproducibility of the configurations in the presence of bimodality, even when the physical observables agree with the theoretical ones. On the other hand, independent of the phase of the system, the information embodied in the neural network weights is sufficient to discriminate whether the configurations come from a given temperature well. The learned representations of the RBM can discriminate system configurations at different temperatures, promising interesting applications in real systems that could help recognize crossover phenomena.

Performance Analysis of Embedded Multilayer Perceptron Artificial Neural Networks on Smart Cyber-Physical Systems for IoT Environments.

At present, modern society is experiencing a significant transformation. Thanks to the digitization of society and manufacturing, mainly because of a combination of technologies, such as the Internet of Things, cloud computing, machine learning, smart cyber-physical systems, etc., which are making the smart factory and Industry 4.0 a reality. Currently, most of the intelligence of smart cyber-physical systems is implemented in software. For this reason, in this work, we focused on the artificial intelligence software design of this technology, one of the most complex and critical. This research aimed to study and compare the performance of a multilayer perceptron artificial neural network designed for solving the problem of character recognition in three implementation technologies: personal computers, cloud computing environments, and smart cyber-physical systems. After training and testing the multilayer perceptron, training time and accuracy tests showed each technology has particular characteristics and performance. Nevertheless, the three technologies have a similar performance of 97% accuracy, despite a difference in the training time. The results show that the artificial intelligence embedded in fog technology is a promising alternative for developing smart cyber-physical systems.

How Does Neural Network Model Capacity Affect Photovoltaic Power Prediction? A Study Case.

The use of models capable of forecasting the production of photovoltaic (PV) energy is essential to guarantee the best possible integration of this energy source into traditional distribution grids. Long Short-Term Memory networks (LSTMs) are commonly used for this purpose, but their use may not be the better option due to their great computational complexity and slower inference and training time. Thus, in this work, we seek to evaluate the use of neural networks MLPs (Multilayer Perceptron), Recurrent Neural Networks (RNNs), and LSTMs, for the forecast of 5 min of photovoltaic energy production. Each iteration of the predictions uses the last 120 min of data collected from the PV system (power, irradiation, and PV cell temperature), measured from 2019 to mid-2022 in Maceió (Brazil). In addition, Bayesian hyperparameters optimization was used to obtain the best of each model and compare them on an equal footing. Results showed that the MLP performs satisfactorily, requiring much less time to train and forecast, indicating that they can be a better option when dealing with a very short-term forecast in specific contexts, for example, in systems with little computational resources.

Reconocimiento de rutas biosintéticas para semioquímicos mediante técnicas de aprendizaje de máquina

En este trabajo consideramos 148 semioquímicos reportados para la familia Scarabaeidae, cuya estructura química fue caracterizada empleando un conjunto de 200 descriptores moleculares de cinco clases distintas. La selección de los descriptores más discriminantes se realizó con tres técnicas: análisis de componentes principales, por cada clase de descriptores, bosques aleatorios y Boruta-Shap, aplicados al total de descriptores. A pesar de que las tres técnicas son conceptualmente diferentes, seleccionan un número de descriptores similar de cada clase. Propusimos una combinación de técnicas de aprendizaje de máquina para buscar un patrón estructural en el conjunto de semioquímicos y posteriormente realizar la clasificación de estos. El patrón se estableció a partir de la alta pertenencia de un subconjunto de estos metabolitos a los grupos que fueron obtenidos por un método de agrupamiento basado en lógica difusa, C-means; el patrón descubierto corresponde a las rutas biosintéticas por las cuales se obtienen biológicamente. Esta primera clasificación se corroboró con el empleo de mapas autoorganizados de Kohonen. Para clasificar aquellos semioquímicos cuya pertenencia a una ruta no quedaba claramente definida, construimos dos modelos de perceptrones multicapa, los cuales tuvieron un desempeño aceptable.

In this work we consider 148 semiochemicals reported for the family Scarabaeidae, whose chemical structure was characterized using a set of 200 molecular descriptors from five different classes. The selection of the most discriminating descriptors was carried out with three different techniques: Principal Component Analysis, for each class of descriptors, Random Forests and Boruta-Shap, applied to the total of descriptors. Although the three techniques are conceptually different, they select a similar number of descriptors from each class. We proposed a combination of machine learning techniques to search for a structural pattern in the set of semiochemicals and then perform their classification. The pattern was established from the high belonging of a subset of these metabolites to the groups that were obtained by a grouping method based on fuzzy C-means logic; the discovered pattern corresponds to the biosynthetic pathway by which they are obtained biologically. This first classification was corroborated with Kohonen's self-organizing maps. To classify those semiochemicals whose belonging to a biosynthetic pathway was not clearly defined, we built two models of Multilayer Perceptrons which had an acceptable performance.

Neste trabalho consideramos 148 semioquímicos reportados para a família Scarabaeidae, cuja estrutura química foi caracterizada usando um conjunto de 200 descritores moleculares de 5 classes diferentes. A seleção dos descritores mais discriminantes foi realizada com três técnicas diferentes: Análise de Componentes Principais, para cada classe de descritores, Florestas Aleatórias e Boruta-Shap, aplicadas a todos os descritores. Embora as três técnicas sejam conceitualmente diferentes, elas selecionaram um número semelhante de descritores de cada classe. Nós propusemos uma combinação de técnicas de aprendizado de máquina para buscar um padrão estrutural no conjunto de semioquímicos e então realizar sua classificação. O padrão foi estabelecido a partir da alta pertinência de um subconjunto desses metabólitos aos grupos que foram obtidos por um método de agrupamento baseado em lógica fuzzy, C-means; o padrão descoberto corresponde às rotas biossintéticas pelas quais eles são obtidos biologicamente. Essa primeira classificação foi corroborada com o uso dos mapas auto-organizados de Kohonen. Para classificar os semioquímicos cuja pertença a uma rota não foi claramente definida, construímos dois modelos de Perceptrons Multicamadas que tiveram um desempenho aceitável.

Wavelet packet transform and multilayer perceptron to identify voices with a mild degree of vocal deviation / Transformada Wavelet packet y Perceptrón Multicapa para identificación de voces con grado leve de desvío vocal

Introduction. Laryngeal disorders are characterized by a change in the vibratory pattern of the vocal folds. This disorder may have an organic origin described by anatomical fold modification, or a functional origin caused by vocal abuse or misuse. The most common diagnostic methods are performed by invasive imaging features that cause patient discomfort. In addition, mild voice deviations do not stop the in-dividual from using their voices, which makes it difficult to identify the problem and increases the possibility of complications. Aim. For those reasons, the goal of the present paper was to develop a noninvasive alternative for the identification of voices with a mild degree of vocal deviation ap-plying the Wavelet Packet Transform (WPT) and Multilayer Perceptron (MLP), an Artificial Neural Network (ANN). Methods. A dataset of 74 audio files were used. Shannon energy and entropy mea-sures were extracted using the Daubechies 2 and Symlet 2 families and then the processing step was performed with the MLP ANN. Results. The Symlet 2 family was more efficient in its generalization, obtaining 99.75% and 99.56% accuracy by using Shannon energy and entropy measures, re-spectively. The Daubechies 2 family, however, obtained lower accuracy rates: 91.17% and 70.01%, respectively. Conclusion. The combination of WPT and MLP presented high accuracy for the identification of voices with a mild degree of vocal deviation

ntroducción. Los trastornos laríngeos se caracterizan por un cambio en el patrón vibratorio de los pliegues vocales. Este trastorno puede tener un origen orgánico, descrito como la modificación anatómica de los pliegues vocales, o de origen fun-cional, provocado por abuso o mal uso de la voz. Los métodos de diagnóstico más comunes se realizan mediante procedimientos invasivos que causan malestar al pa-ciente. Además, los desvíos vocales de grado leve no impiden que el individuo utilice la voz, lo que dificulta la identificación del problema y aumenta la posibilidad de complicaciones futuras.Objetivo. Por esas razones, el objetivo de esta investigación es desarrollar una he-rramienta alternativa, no invasiva para la identificación de voces con grado leve de desvío vocal aplicando Transformada Wavelet Packet (WPT) y la red neuronal artifi-cial del tipo Perceptrón Mutlicapa (PMC). Métodos. Fue utilizado un banco de datos con 78 voces. Fueron extraídas las me-didas de energía y entropía de Shannon usando las familias Daubechies 2 y Symlet 2 para después aplicar la red neuronal PMC. Resultados. La familia Symlet 2 fue más eficiente en su generalización, obteniendo un 99.75% y un 99.56% de precisión mediante el uso de medidas de energía y en-tropía de Shannon, respectivamente. La familia Daubechies 2, sin embargo, obtuvo menores índices de precisión: 91.17% y 70.01%, respectivamente. Conclusión. La combinación de WPT y PMC presentó alta precisión para la iden-tificación de voces con grado leve de desvío vocal

Raman Spectroscopy and Machine-Learning for Early Detection of Bacterial Canker of Tomato: The Asymptomatic Disease Condition.

Bacterial canker of tomato is caused by Clavibacter michiganensis subsp. michiganensis (Cmm). The disease is highly destructive, because it produces latent asymptomatic infections that favor contagion rates. The present research aims consisted on the implementation of Raman spectroscopy (RS) and machine-learning spectral analysis as a method for the early disease detection. Raman spectra were obtained from infected asymptomatic tomato plants (BCTo) and healthy controls (HTo) with 785 nm excitation laser micro-Raman spectrometer. Spectral data were normalized and processed by principal component analysis (PCA), then the classifiers algorithms multilayer perceptron (PCA + MLP) and linear discriminant analysis (PCA + LDA) were implemented. Bacterial isolation and identification (16S rRNA gene sequencing) were realized of each plant studied. The Raman spectra obtained from tomato leaf samples of HTo and BCTo exhibited peaks associated to cellular components, and the most prominent vibrational bands were assigned to carbohydrates, carotenoids, chlorophyll, and phenolic compounds. Biochemical changes were also detectable in the Raman spectral patterns. Raman bands associated with triterpenoids and flavonoids compounds can be considered as indicators of Cmm infection during the asymptomatic stage. RS is an efficient, fast and reliable technology to differentiate the tomato health condition (BCTo or HTo). The analytical method showed high performance values of sensitivity, specificity and accuracy, among others.

Artificial intelligence in the IVF laboratory: overview through the application of different types of algorithms for the classification of reproductive data.

Over the past years, the assisted reproductive technologies (ARTs) have been accompanied by constant innovations. For instance, intracytoplasmic sperm injection (ICSI), time-lapse monitoring of the embryonic morphokinetics, and PGS are innovative techniques that increased the success of the ART. In the same trend, the use of artificial intelligence (AI) techniques is being intensively researched whether in the embryo or spermatozoa selection. Despite several studies already published, the use of AI within assisted reproduction clinics is not yet a reality. This is largely due to the different AI techniques that are being proposed to be used in the daily routine of the clinics, which causes some uncertainty in their use. To shed light on this complex scenario, this review briefly describes some of the most frequently used AI algorithms, their functionalities, and their potential use. Several databases were analyzed in search of articles where applied artificial intelligence algorithms were used on reproductive data. Our focus was on the classification of embryonic cells and semen samples. Of a total of 124 articles analyzed, 32 were selected for this review. From the proposed algorithms, most have achieved a satisfactory precision, demonstrating the potential of a wide range of AI techniques. However, the evaluation of these studies suggests the need for more standardized research to validate the proposed models and their algorithms. Routine use of AI in assisted reproduction clinics is just a matter of time. However, the choice of AI technique to be used is supported by a better understanding of the principles subjacent to each technique, that is, its robustness, pros, and cons. We provide some current (although incipient) and potential uses of AI on the clinic routine, discussing how accurate and friendly it could be. Finally, we propose some standards for AI research on the selection of the embryo to be transferred and other future hints. For us, the imminence of its use is evident, providing a revolutionary milestone that will impact the ART.

Attribute selection using correlations and principal components for artificial neural networks employment for landslide susceptibility assessment.

Landslide susceptibility maps can be developed with artificial neural networks (ANNs). In order to train our ANNs, a digital elevation model (DEM) and a scar map of one previous event were used. Eleven attributes are generated, possibly containing redundant information. Our base model is formed by, essentially, one input (the DEM), eleven attributes, 30 neurons, and one output (susceptibility). Principal components (PCs) group information in the first projected variables, the last ones can be expendable. In the present paper, four groups of models were trained: one with eleven attributes generated from the DEM; one with 8 out of 11 attributes, in which 3 were eliminated by their high correlation with others; other, with the data projected over its PCs; and another, using 8 out of 11 PCs. The used number of neurons in hidden layer is 30, calibrated based on a complexity analysis that is an in-house developed method. The ANN models trained with the original data generated better statistical results than their counterparts trained with the PC transformed input. Keeping the original 11 attributes calculated provided the best metrics among all models, showing that eliminating attributes also eliminates information used by the model. Using 11 PC transformed attributes hindered trained. However, for the model with eight PCs, training was much faster than its counterpart with little accuracy loss. The metrics and maps achieved were considered acceptable, conveying the power of our model based on ANNs, which uses essentially one input (the DEM) for mapping areas susceptible to mass movements.

Objective Assessment of Skin Repigmentation Using a Multilayer Perceptron.

BACKGROUND: Vitiligo is a pathology that causes the appearance of achromic macules on the skin that can spread on to other areas of the body. It is estimated that it affects 1.2% of the world population and can disrupt the mental state of people in whom this disease has developed, generating negative feelings that can become suicidal in the worst of cases. The present work focuses on the development of a support tool that allows to objectively quantifying the repigmentation of the skin. METHODS: We propose a novel method based on artificial neural networks that use characteristics of the interaction of light with the skin to determine areas of healthy skin and skin with vitiligo. We used photographs of specific areas of skin containing vitiligo. We select as independent variables: the type of skin, the amount of skin with vitiligo and the amount of repigmented skin. Considering these variables, the experiments were organized in an orthogonal table. We analyzed the result of the method based on three parameters (sensitivity, specificity, and F1-Score) and finally, its results were compared with other methods proposed in similar research. RESULTS: The proposed method demonstrated the best performance of the three methods, and it also showed its capability to detect healthy skin and skin with vitiligo in areas up to 1 × 1 pixels. CONCLUSION: The results show that the proposed method has the potential to be used in clinical applications. It should be noted that the performance could be significantly improved by increasing the training patterns.

Neural networks in spatialization of meteorological elements and their application in the climatic agricultural zoning of bamboo.

Bamboo has an important role in international commerce due to its diverse uses, but few studies have been conducted to evaluate its climatic adaptability. Thus, the objective of this study was to construct an agricultural zoning for climate risk (ZARC) for bamboo using meteorological elements spatialized by neural networks. Climate data included air temperature (TAIR, °C) and rainfall (P) from 4947 meteorological stations in Brazil from the years 1950 to 2016. Regions were considered climatically apt for bamboo cultivation when TAIR varied between 18 and 35 °C, and P was between 500 and 2800 mm year-1, or PWINTER was between 90 and 180 mm year-1. The remainder of the areas was considered marginal or inapt for bamboo cultivation. A multilayer perceptron (MLP) neural network with a multilayered "backpropagation" training algorithm was used to spatialize the territorial variability of each climatic element for the whole area of Brazil. Using the overlapping of the TAIR, P, and PWINTER maps prepared by MLP, and the established climatic criteria of bamboo, we established the agricultural zoning for bamboo. Brazil demonstrates high seasonal climatic variability with TAIR varying between 14 and 30 °C, and P varying between < 400 and 4000 mm year-1. The ZARC showed that 87% of Brazil is climatically apt for bamboo cultivation. The states that were classified as apt in 100% of their territories were Mato Grosso do Sul, Goiás, Tocantins, Rio de Janeiro, Espírito Santo, Sergipe, Alagoas, Ceará, Piauí, Maranhão, Rondônia, and Acre. The regions that have restrictions due to low TAIR represent just 11% of Brazilian territory. This agroclimatic zoning allowed for the classification of regions based on aptitude of climate for bamboo cultivation and showed that 71% of the total national territory is considered to be apt for bamboo cultivation. The regions that have restrictions are part of southern Brazil due to low values of TAIR and portions of the northern region that have high levels of P which is favorable for the development of diseases.

Tuberculosis diagnosis support analysis for precarious health information systems.

BACKGROUND AND OBJECTIVE: Pulmonary tuberculosis is a world emergency for the World Health Organization. Techniques and new diagnosis tools are important to battle this bacterial infection. There have been many advances in all those fields, but in developing countries such as Colombia, where the resources and infrastructure are limited, new fast and less expensive strategies are increasingly needed. Artificial neural networks are computational intelligence techniques that can be used in this kind of problems and offer additional support in the tuberculosis diagnosis process, providing a tool to medical staff to make decisions about management of subjects under suspicious of tuberculosis. MATERIALS AND METHODS: A database extracted from 105 subjects with precarious information of people under suspect of pulmonary tuberculosis was used in this study. Data extracted from sex, age, diabetes, homeless, AIDS status and a variable with clinical knowledge from the medical personnel were used. Models based on artificial neural networks were used, exploring supervised learning to detect the disease. Unsupervised learning was used to create three risk groups based on available information. RESULTS: Obtained results are comparable with traditional techniques for detection of tuberculosis, showing advantages such as fast and low implementation costs. Sensitivity of 97% and specificity of 71% where achieved. CONCLUSIONS: Used techniques allowed to obtain valuable information that can be useful for physicians who treat the disease in decision making processes, especially under limited infrastructure and data.

Fangorn Forest (F2): a machine learning approach to classify genes and genera in the family Geminiviridae.

BACKGROUND: Geminiviruses infect a broad range of cultivated and non-cultivated plants, causing significant economic losses worldwide. The studies of the diversity of species, taxonomy, mechanisms of evolution, geographic distribution, and mechanisms of interaction of these pathogens with the host have greatly increased in recent years. Furthermore, the use of rolling circle amplification (RCA) and advanced metagenomics approaches have enabled the elucidation of viromes and the identification of many viral agents in a large number of plant species. As a result, determining the nomenclature and taxonomically classifying geminiviruses turned into complex tasks. In addition, the gene responsible for viral replication (particularly, the viruses belonging to the genus Mastrevirus) may be spliced due to the use of the transcriptional/splicing machinery in the host cells. However, the current tools have limitations concerning the identification of introns. RESULTS: This study proposes a new method, designated Fangorn Forest (F2), based on machine learning approaches to classify genera using an ab initio approach, i.e., using only the genomic sequence, as well as to predict and classify genes in the family Geminiviridae. In this investigation, nine genera of the family Geminiviridae and their related satellite DNAs were selected. We obtained two training sets, one for genus classification, containing attributes extracted from the complete genome of geminiviruses, while the other was made up to classify geminivirus genes, containing attributes extracted from ORFs taken from the complete genomes cited above. Three ML algorithms were applied on those datasets to build the predictive models: support vector machines, using the sequential minimal optimization training approach, random forest (RF), and multilayer perceptron. RF demonstrated a very high predictive power, achieving 0.966, 0.964, and 0.995 of precision, recall, and area under the curve (AUC), respectively, for genus classification. For gene classification, RF could reach 0.983, 0.983, and 0.998 of precision, recall, and AUC, respectively. CONCLUSIONS: Therefore, Fangorn Forest is proven to be an efficient method for classifying genera of the family Geminiviridae with high precision and effective gene prediction and classification. The method is freely accessible at www.geminivirus.org:8080/geminivirusdw/discoveryGeminivirus.jsp .

Neuro-genetic system for optimization of GMI samples sensitivity.

Magnetic sensors are largely used in several engineering areas. Among them, magnetic sensors based on the Giant Magnetoimpedance (GMI) effect are a new family of magnetic sensing devices that have a huge potential for applications involving measurements of ultra-weak magnetic fields. The sensitivity of magnetometers is directly associated with the sensitivity of their sensing elements. The GMI effect is characterized by a large variation of the impedance (magnitude and phase) of a ferromagnetic sample, when subjected to a magnetic field. Recent studies have shown that phase-based GMI magnetometers have the potential to increase the sensitivity by about 100 times. The sensitivity of GMI samples depends on several parameters, such as sample length, external magnetic field, DC level and frequency of the excitation current. However, this dependency is yet to be sufficiently well-modeled in quantitative terms. So, the search for the set of parameters that optimizes the samples sensitivity is usually empirical and very time consuming. This paper deals with this problem by proposing a new neuro-genetic system aimed at maximizing the impedance phase sensitivity of GMI samples. A Multi-Layer Perceptron (MLP) Neural Network is used to model the impedance phase and a Genetic Algorithm uses the information provided by the neural network to determine which set of parameters maximizes the impedance phase sensitivity. The results obtained with a data set composed of four different GMI sample lengths demonstrate that the neuro-genetic system is able to correctly and automatically determine the set of conditioning parameters responsible for maximizing their phase sensitivities.

Odorant recognition using biological responses recorded in olfactory bulb of rats.

In this study we applied pattern recognition (PR) techniques to extract odorant information from local field potential (LFP) signals recorded in the olfactory bulb (OB) of rats subjected to different odorant stimuli. We claim that LFP signals registered on the OB, the first stage of olfactory processing, are stimulus specific in animals with normal sensory experience, and that these patterns correspond to the neural substrate likely required for perceptual discrimination. Thus, these signals can be used as input to an artificial odorant classification system with great success. In this paper we have designed and compared the performance of several configurations of artificial olfaction systems (AOS) based on the combination of four feature extraction (FE) methods (Principal Component Analysis (PCA), Fisher Transformation (FT), Sammon NonLinear Map (NLM) and Wavelet Transform (WT)), and three PR techniques (Linear Discriminant Analysis (LDA), Multilayer Perceptron (MLP) and Support Vector Machine (SVM)), when four different stimuli are presented to rats. The best results were reached when PCA extraction followed by SVM as classifier were used, obtaining a classification accuracy of over 95% for all four stimuli.

Lukasiewicz logic based Fuzzy similarity classifier for Denver group chromosomal classification / Utilizando lógica de Lukasiewicz baseada em similaridade Fuzzy na classificação de cromossomos do tipo Denver group

This paper proposes a novel P1-weighted Lukasiewicz Logic based Fuzzy Similarity Classifier for classifying Denver Group of chromosomes and compares its performance with the other classifiers under study. A chromosome is classified to one of the seven groups from A to G, based on the Denver System of classification of chromosomes. Chromosomes within a particular Denver Group are difficult to identify, possessing almost identical characteristics for the extracted features. This work evaluates the performance of supervised classifiers including Naive Bayes, Support Vector Machine with Gaussian Kernel (SVM), Multilayer perceptron (MLP) and a novel, unsupervised, P1-weighted Lukasiewicz Logic based Fuzzy Similarity Classifier, in classifying the Denver Group of chromosomes. A fundamental review on fuzzy similarity based classification is presented. Experimental results clearly demonstrates that the proposed P1-weighted Lukasiewicz Logic based Fuzzy Similarity Classifier using the generalized Minkowski mean metric, produces the best classification results, almost identical to the Ground Truth values. One-way Analysis of Variance (ANOVA) at 95% and 99% level of confidence and Tukey's post-hoc analysis is performed to validate the selection of the classifier. The proposed P1-weighted Lukasiewicz Logic based Fuzzy Similarity Classifier gives the most promising classification results and can be applied to any large scale biomedical data and other applications.

Este trabalho propõe uma nova lógica P1pondera de Lukasiewicz de acordo com o classificador de similarida fuzzy para classificar cromossomas do Grupo Denver e compara o seu desempenho com os outros classificadores em estudo. Um cromossoma é classificado com um dos sete grupos de A a G, com base no Sistema de Denver de classificação de cromossomos. Cromossomos dentro de um grupo de Denver particular são difíceis de identificar, com características quase idênticas para os recursos extraídos. Este trabalho avalia o desempenho de classificadores supervisionados, incluindo Naive Bayes, Support Vector Machine com Gaussian Kernel (SVM), perceptron multicamadas (MLP) e um novo classificador sem supervisão, P1-weighted, lógica de Lukasiewicz de acordo com o classificador de similaridade Fuzzy para a classificação do Grupo Denver de cromossomos . Apresenta-se ma revisão fundamentada na classificação de acordo com similaridade difusa. Resultados experimentais demonstram claramente que Classificador Similaridade Fuzzy proposto de acordo com a lógica de Lukasiewicz P1-weighted usando a médica métrica de Minkowski para produz melhores resultados de classificação. Estes valores foram muito similares aos valores de Ground Truth . Análise de variancia (ANOVA) com 95% de grau de confiança e análise post-hoc de Tukey 99% foram realizadas para validar a seleção do classificador. Este classificador P1-weighted de lógica de Lukasiewicz está de acordo com o classificador de similaridade difusa oferecendo resultados declassificação mais promissoras. Portanto, podendo ser aplicado a dados biomédicos em larga escala além de outras aplicações.

An artificial neural network model for the prediction of mechanical and barrier properties of biodegradable films.

Nowadays, the production of biodegradable starch-based films is of great interest because of the growing environmental concerns regarding pollution and the need to reduce dependence on the plastics industry. A broad view of the role of different components, added to starch-based films to improve their properties, is required to guide the future development. The self-organizing maps (SOMs) provide comparisons that initially were complicated due to the large volume of the data. Furthermore, the construction of a model capable of predicting the mechanical and barrier properties of these films will accelerate the development of films with improved characteristics. The water vapor permeability (WVP) analysis using the SOM algorithm showed that the presence of glycerol is very important for films with low amounts of poly (butylene adipate co-terephthalate) and confirms the role of the equilibrium relative humidity in the determination of WVP. Considering the mechanical properties, the SOM analysis emphasizes the important role of poly (butylene adipate co-terephthalate) in thermoplastic starch based films. The properties of biodegradable films were predicted and optimized by using a multilayer perceptron coupled with a genetic algorithm, presenting a great correlation between the experimental and theoretical values with a maximum error of 24%. To improve the response of the model and to ensure the compatibility of the components more information will be necessary.

Modelación de l estructura jerárquica de macronvertebrados bentónicos a travéés de redes neuronales artificiales / Modeling of the Hierarchical Structure of Freshwater Macroinvertebrates Using Artificial Neural Networks

El estudio de la estructura jerárquica de comunidades ecológicas, se ha sintetizado de manera regular a través de técnicas multivariadas de ordenación o clasificación. Sin embargo, al contarse actualmente con herramientas analíticas de computación bioinspirada provenientes de la inteligencia artificial, existe la oportunidad de establecer modelos ecológicos, con características deseables como flexibilidad, exactitud, robustez y confiabilidad. En este contexto, esta investigación utilizó dos métodos computacionales de utilidad en ecoinformática, referidos a redes neuronales artificiales (RNARs) para la modelación de la estructura jerárquica de una comunidad de macroinvertebrados bentónicos en términos de auto-organización y predicción. El primer método de modelación consistió en un mapa de auto-organización (MAU), una herramienta de aprendizaje no supervisado que clasificó las especies de macroinvertebrados; este MAU tomó en la capa de entrada la abundancia de cada taxa, y en la de salida proyectó su clasificación en 15 unidades y cuatro agrupamientos jerárquicos. La segunda RNA, correspondió a un Perceptrón multicapa de alimentación adelantada con algoritmo de retropropagación, que modeló separadamente la riqueza y la abundancia de Ephemeroptera, Coleoptera y Trichoptera (ECT), en función de nueve variables fisicoquímicas; la arquitectura del perceptrón correspondió a una constitución de nueve, siete, y una neurona en las capas de entrada, intermedia y salida, respectivamente. Los resultados sugieren que las RNARs utilizadas evidenciaron tanto los patrones jerárquicos, como los de riqueza y abundancia de ECT de manera adecuada, al tiempo que facilitaron el análisis de los datos y el entendimiento de la dinámica de la comunidad de macroinvertebrados, objeto de estudio.

The study of hierarchical structures of ecological communities has been synthesized in an ordinary way by means of multivariated techniques of ordination or clustering. Currently, analytical tools of bio-inspired computation belonging to the area of artificial intelligence are available to achieve ecological models with desirable characteristics, such as; flexibility, accuracy, robustness and reliability. In this context, this study employed two computational methods useful in ecoinformatics referring to artificial neural networks (RNAR) for the modeling of the hierarchical structure of a benthic macroinvertebrate community in self-organization and prediction terms. The first ANN modeling method consisted of a Kohonen self-organization map (SOM), a non-supervised learning tool that classify the species of macroinvertebrates; this SOM in the input layer of gets the abundance of each ‘taxa’ from the data matrix, while in the output layer was visualized the computational results. Thus, in the output layer the species are organized in fifteen units and four hierarchical clusters. The second ANN method applied consisted of a multilayer feed-forward perceptron net with back-propagation algorithm to predict the three major insect orders; this means, Ephemeroptera, Coleoptera and Trichoptera (ECT) richness and abundance using a set of nine physical-chemical variables. This ANN architecture included a neuron for each environmental variable, a hidden layer with seven neurons and a neuron in the output layer for ECT prediction. The results suggest that both types of ANN used, SOM and perceptron, were correspondingly related to the hierarchical patterns and with the richness and abundance patterns’ predictions, and gave the data analysis and understanding of the dynamic of the macroinvertebrates community, in a correct way.