Combining deep learning and X-ray imaging technology to assess tomato seed quality

Traditional germination tests which assess seed quality are costly and time-consuming, mainly when performed on a large scale. In this study, we assessed the efficiency of X-ray imaging analyses in predicting the physiological quality of tomato seeds. A convolutional neural network (CNN) called mask region convolutional neural network (MaskRCNN) was also tested for its precision in adequately classifying tomato seeds into four seed quality categories. For this purpose, X-ray images were taken of seeds of 49 tomato genotypes (46 Solanum pennellii introgression lines) from two different growing seasons. Four replicates of 25 seeds for each genotype were analyzed. These seeds were further assessed for germination and seedling vigor-related traits in two independent trials. Correlation analysis revealed significant linear association between germination and image-based variables. Most genotypes differed in terms of germination and seed development performance considering the two independent trials, except LA 4046, LA 4043, and LA4047, which showed similar behavior. Our findings point out that seeds with low opacity and percentage of damaged seed tissue and high values for living tissue opacity have greater physiological quality. In short, our work confirms the reliability of X-ray imaging and deep learning methodologies in predicting the physiological quality of tomato seeds.

Artificial neural networks for the management of poultry industry: a simulation based on the broiler production chain / Redes neurais artificiais para o gerenciamento da indústria avícola: uma simulação baseada na cadeia de produção de frangos de corte

The aim of this study was to predict production indicators and to determine their potential economic impact on a poultry integration system using artificial neural networks (ANN) models. Forty zootechnical and production parameters from broiler breeder farms, one hatchery, broiler production flocks, and one slaughterhouse were selected as variables. The ANN models were established for four output variables: "saleable hatching", "weight at the end of week 5," "partial condemnation," and "total condemnation" and were analyzed in relation to the coefficient of multiple determination (R2), correlation coefficient (R), mean error (E), mean squared error (MSE), and root mean square error (RMSE). The production scenarios were simulated and the economic impacts were estimated. The ANN models were suitable for simulating production scenarios after validation. For "saleable hatching", incubator and egg storage period are likely to increase the financial gains. For "weight at the end of the week 5" the lineage (A) is important to increase revenues. However, broiler weight at the end of the first week may not have a significant influence. Flock sex (female) may influence the "partial condemnation" rates, while chick weight at first day may not. For "total condemnation", flock sex and type of chick may not influence condemnation rates, but mortality rates and broiler weight may have a significant impact.

O objetivo deste trabalho foi predizer os indicadores de produção e determinar o seu potencial impacto econômico em um sistema de integração utilizando as redes neurais artificiais (RNA). Quarenta parâmetros zootécnicos e de produção de granjas de matrizes e de frango de corte, um incubatório e um abatedouro foram selecionados como variáveis. Os modelos de RNA foram estabelecidos para quatro variáveis de saída ("eclosão vendável", "peso ao final da quinta semana", "condenações parciais" e "condenações totais") e foram analisados em relação ao coeficiente de determinação múltipla (R2), coeficiente de correlação (R), erro médio (E), erro quadrático médio (EQM) e raiz do erro quadrático médio (REQM). Os cenários produtivos foram simulados e os impactos foram estimados. Os modelos de RNA gerados foram adequados para simular diferentes cenários produtivos após o treinamento. Para "eclosão vendável", o modelo de incubadora e o período de incubação aumentaram os ganhos financeiros. Para "peso ao final da quinta semana", a linhagem também demonstrou influencia no retorno financeiro, o que não aconteceu com o peso ao final da primeira semana. O sexo do lote possui influência nas taxas de "condenação parcial", ao contrário do peso do frango no primeiro dia. As taxas de mortalidade e o peso do frango apresentaram influência na "condenação total", mas o sexo do lote e o tipo de pinto não tiverem influência.

Developing a hybrid forecasting system for agricultural commodity prices (case study: Thailand rice free on board price) / Desenvolvimento de um sistema de previsão híbrido para preços de commodities agrícolas: estudo de caso para arroz tailandês

Forecast the price of agricultural goods is a beneficial action for farmers, marketing agents, consumers, and policymakers. Today, managing this product security requires price forecasting models that are both efficient and reliable for a country's import and export. In the last few decades, the Autoregressive Integrated Moving Average (ARIMA) model has been widely used in economics time series forecasting. Recently, many of the time series observations presented in economics have been clearly shown to be nonlinear, Machine learning (ML) modelling, conversely, offers a potential price forecasting technique that is more flexible given the limited data available in most countries' economies. In this research, a hybrid price forecasting model has been used, through a novel clustering technique, a new cluster selection algorithm and a multilayer perceptron neural network (MLPNN), which had many advantages and using monthly time series of Thai rice FOB price form November 1987 to October 2017. The empirical results of this study showed that the value of root mean square error (RMSE) equals 14.37 and the Mean absolute percentage error (MAPE) equals 4.09% for the hybrid model. The evaluation results of proposed method and comparison its performance with four benchmark models, by monthly time series of Thailand rice FOB price from November 1987 to October 2017 showed the outperform of proposed method.

Prever o preço dos produtos agrícolas é uma ação benéfica para agricultores, agentes de marketing, consumidores e legisladores. Hoje, o gerenciamento da segurança desse produto requer modelos de previsão de preços eficientes e confiáveis para a importação e exportação de um país. Nas últimas décadas, o modelo Autoregressive Integrated Moving Average (ARIMA) tem sido amplamente utilizado na previsão de séries temporais da economia. Recentemente, muitas das observações de séries temporais apresentadas em economia têm se mostrado claramente não lineares. A modelagem de aprendizado de máquina (ML), por outro lado, oferece uma técnica de previsão de preços potencial que é mais flexível, apresentados os dados limitados disponíveis na maioria dos países. Nesta pesquisa, um modelo híbrido de previsão de preços foi usado, por meio de uma nova técnica de agrupamento, um novo algoritmo de seleção de agrupamento e uma rede neural perceptron multicamadas (MLPNN), que teve muitas vantagens, e usando séries temporais mensais de preços FOB do arroz tailandês de novembro 1987 a outubro de 2017. Os resultados empíricos deste estudo mostraram que o valor da raiz do erro quadrático médio (RMSE) é igual a 14,37 e o erro percentual absoluto médio (MAPE) é igual a 4,09% para o modelo híbrido. Os resultados da avaliação do método proposto e a comparação de seu desempenho com quatro modelos de benchmark, por séries temporais mensais de preço FOB do arroz tailandês de novembro de 1987 a outubro de 2017, mostram o desempenho superior do método proposto.

Artificial neural networks in the prediction of fraud in integral milk powder by adding whey powder / Redes neurais artificiais na predição de fraude em leite em pó integral pela adição de soro lácteo em pó

This research was performed to ascertain the most suitable Artificial Neural Network (ANN) model to quantify the degree of fraud in powdered milk through the addition of powdered whey via regular standard physicochemical analyses. In this study, an evaluation was done on 103 samples with different quantities of added whey powder to whole milk powder. Using Fourier Transform Infrared Spectroscopy the fat, cryoscopy, total solids, defatted dry extract, lactose, protein and casein were analyzed. The hyperbolic tangent transformation function was used with 45 topologies, and the Holdback and K-fold validation methods were tested. In the Holdback method, 75% of the database was employed for training, while 25% was used for validation. In the K-fold method, the database was categorized into five equal sized subsets, which alternated between training and validation. Of the two methods, the K-fold method was proven to have superior efficiency. Next, analysis was done on three models of multilayer perceptron networks with feedforward architecture. In Model 1, the input layer contained all the physicochemical analyses conducted, in model 2 the casein analysis was excluded, and in model 3 the routine analyses performed for dairy products was done (fat, defatted dry extract, cryoscopy and total solids). From Model 3 an ANN was derived which could satisfactorily predict fraud calculated from using the routine and standard analyses for dairy products, containing 64 nodes in the hidden layer, with R² of 0.9935 and RMSE of 0.5779 for training, and R² of 0.9964 and RMSE of 0.4358 for validation.

O objetivo do trabalho foi determinar o melhor modelo de rede neural artificial (RNA) para quantificar fraude em leite em pó, pela adição de soro em pó, por meio de analises físico-químicas de rotina. Foram avaliados 103 níveis de adição de soro lácteo em pó em leite em pó integral. As análises de gordura, crioscopia, sólidos totais, extrato seco desengordurado, lactose, proteína e caseína foram realizadas por espectroscopia no infravermelho com transformada de Fourier. A função de transformação utilizada foi a tangente hiperbólica, em que testou-se 45 topologias e dois métodos de validação: holdback e k-fold. Para o método holdback, 75% do banco de dados foi utilizado para o treinamento e 25% para a validação. Para o método k-fold, o banco de dados foi dividido em cinco subconjuntos de mesmo tamanho que se alternavam entre treinamento e validação. O método k-fold se mostrou mais eficiente. Três modelos de redes perceptron de múltiplas camadas com arquitetura feedforward foram analisados. No modelo 1 a camada de entrada constituía todas as análises físico-químicas realizadas, no modelo 2 excluiu-se a análise de caseína e no modelo 3 utilizou-se as análises de rotina em laticínios (gordura, extrato seco desengordurado, crioscopia e sólidos totais). O modelo 3 obteve uma RNA capaz de predizer satisfatoriamente a fraude avaliada a partir de análises consideradas de rotina em laticínios com uma RNA contendo 64 nodos na camada oculta, R² de 0,9935 e RMSE de 0,5779 para treinamento, R² de 0,9964 e RMSE de 0,4358 para validação.

Artificial Neural Networks to Predict Egg-Production Traits in Commercial Laying Breeder Hens

In recent years, egg production has had an intense growth in Brazil, and Brazilian egg consumption per capita has significantly increased in the last decade. To reduce sanitary and financial risks, decisions regarding the production and health status of the flock must be made based on objective criteria. Our aim was to determine the main "input" variables for the prediction of egg production performance in commercial laying breeder flocks using an ANN model. The software NeuroShellClassifier and NeuroShell Predictor were used to build the ANN. A total of 26 egg-production traits were selected as input variables and eight as output variables. A database of 44,120 Excel cells was generated. For the training and validation of the models, 74.9% and 25.1% of the data were used, respectively. The accuracy of the ANN models was calculated and compared using the analysis of coefficient of multiple determination (R2), mean squared error (MSE), and an assessment of uniform scatter in the residual plots. The models for the outputs "weekly egg production," "weekly incubated egg,", "accumulated commercial egg," and "viability" showed an R2 greater than 0.8. Other models yielded R2 values lower than 0.8. The ANN predicts adequately eight egg-production traits in the breeders of commercial laying hens. The method is an option for data management analysis in the egg industry, providing estimates of the relative contribution of each input variable to the outputs.(AU)

Multilayer perceptron applied to genotypes classification in diallel studies

In the last decades, a new trend to use more refined analytical procedures, such as artificial neural networks (ANN), has emerged to be most accurate, efficient, and extensively applied for mining and data prediction in different contexts, including plant breeding. Thus, this study was developed to establish a new classification proposal for targeting genotypes in breeding programs to approach classical models, such as a complete diallel and modern prediction techniques. The study was based on the standard deviation values of an interpopulation diallel and it also verified the possibility of training a neural network with the standardized genetic parameters for a discrete scale. We used 12 intercrossed maize populations in a complete diallel scheme (66 hybrids), evaluated during the 2005/2006 crop season in three different environments in southern Brazil. The implemented MLP architecture and other associated parameters allowed the development of a generalist model of genotype classification. The MLP neural network model was efficient in predicting parental and interpopulation hybrid classifications from average genetic components from a complete diallel, regardless of the evaluation environment.(AU)

Machine learning applied to the prediction of root architecture of soybean cultivars under two water availability conditions / Machine learning aplicado à predição da arquitetura radicular de cultivares de soja sob duas condições de disponibilidade hídrica

The objective of this study was to evaluate the performance of four machine learning models, as well as multitask learning, to predict soybean root variables from simpler variables, under two water availability conditions. In order to do so, 100 soybean cultivars were conducted in a greenhouse under a control condition and a stress condition. Aerial part and root variables were evaluated. The machine learning models used to predict complex root variables were artificial neural network (ANN), random forest (RF), extreme gradient boosting (EGBoost) and support vector machine (SVM). A linear model was used for comparison purposes. Multitask learning was employed for ANN and RF. In addition, feature importance was defined using RF and XGBoost algorithms. All the machine learning models performed better than the linear model. In general, SVM had the greatest potential for the prediction of most of the root variables, with better values of RMSE, MAE and R2. Dry weight of the aerial part and root volume exhibited the greatest importance in the predictions. The models developed using multitask learning performed similarly to the ones conventionally developed. Finally, it is concluded that the machine learning models evaluated can be used to predict root variables of soybean from easily measurable variables, such as dry weight of the aerial part and root volume.(AU)

O objetivo deste estudo foi avaliar o desempenho de quatro modelos de machine learning, bem como multitask learning, para predizer variáveis radiculares de soja a partir de variáveis simples, em duas condições de disponibilidade hídrica. Para isso,100 cultivares de soja foram conduzidas em casa de vegetação sob uma condição controle e uma condição estresse. Foram avaliadas as variáveis da parte aérea e da raiz. Os modelos machine learning usados para predizer variáveis complexas do sistema radicular foram rede neural artificial (RNA), random forest (RF), extreme gradient boosting (EGBoost) e support vector machine (SVM). O modelo linear foi usado para fins de comparação. O multitask learning foi empregado para RNA e RF. Além disso, a importância das variáveis foi definida usando algoritmos RF e XGBoost. Todos os modelos de machine learning apresentaram melhor desempenho do que o modelo linear. Em geral, SVM apresentou o maior potencial de predição da maioria das variáveis raiz, com melhores valores de RMSE, MAE e R2. O peso seco da parte aérea e o volume da raiz exibiram as maiores importâncias nas predições. Os modelos desenvolvidos por meio do multitask learning apresentaram desempenhos semelhantes aos desenvolvidos convencionalmente. Por fim, conclui-se que os modelos de machine learning avaliados podem ser usados para predizer variáveis radiculares de soja a partir de variáveis facilmente mensuráveis, como massa seca da parte aérea e volume radicular.(AU)

Evaluation of automated monitoring calving prediction in dairy buffaloes a new tool for calving management / Avaliação do monitoramento automatizado da previsão de parto em búfalos leiteiros: uma nova ferramenta para o gerenciamento de parto

Buffalo is one of the leading milk-producing dairy animals. Its production and reproduction are affected due to some factors including inadequate monitoring around parturition, which cause economic losses like delayed birth process, increased risk of stillbirth, etc. The appropriate calving monitoring is essential for dairy herd management. Therefore, we designed a study its aim was, to predict the calving based on automated machine measured prepartum behaviors in buffaloes. The data were collected from n=40 pregnant buffaloes of 2nd to 5th parity, which was synchronized. The NEDAP neck and leg logger tag was attached to each buffalo at 30 days before calving and automatically collected feeding, rumination, lying, standing, no. of steps, no. of switches from standing to lying (lying bouts) and total motion activity. All behavioral data were reduced to -10 days before the calving date for statistical analysis to use mixed model procedure and ANOVA. Results showed that feeding and rumination time significantly (P<0.05) decreased from -10 to -1 days before calving indicating calving prediction. Moreover, Rumination time was at lowest (P<0.001) value at 2h before the calving such behavioral changes may be useful to predict calving in buffaloes. Similarly, lying bouts and standing time abruptly decreased (P<0.05) from -3 to -1 days before calving, while lying time abruptly increased (P<0.01) from -3 to -1 days before calving (531.57±23.65 to 665.62±18.14, respectively). No. of steps taken and total motion significantly (P<0.05) increased from -10 to -1 days before calving. Feeding time was significantly (P<0.02) lowered in 3rd parity buffaloes compared with 2nd, 4th and 5th parity buffaloes, while standing time of 5th parity buffaloes were lowered (P<0.05) as compared to 2nd to 4th parity buffalos at -1 day of prepartum. However, rumination, lying, no. of steps taken and total motion activity at -1 day of prepartum was independent (P>0.05) of parity in buffaloes. Neural network analysis for combined variables from NEDAP technology at the daily level yielded 100.0% sensitivity and 98% specificity. In conclusion NEDAP technology can be used to measured behavioral changes -10 day before calving as it can serve as a useful guide in the prediction calving date in the buffaloes.(AU)

O búfalo é um dos principais animais produtores de leite. Sua produção e sua reprodução são afetadas por causa de alguns fatores, incluindo o monitoramento inadequado ao redor do parto, que causam perdas econômicas, como atraso no processo de parto, aumento do risco de natimorto, etc. O monitoramento adequado do parto é essencial para o manejo do rebanho leiteiro. Portanto, projetamos um estudo cujo objetivo foi prever o parto com base em comportamentos pré-parto medidos por máquina automatizada em búfalas. Os dados foram coletados de 40 búfalas prenhes de 2ª a 5ª paridade, que foi sincronizada. A etiqueta NEDAP de pescoço e perna foi fixada em cada búfala 30 dias antes do parto e coletava dados, automaticamente, durante a alimentação e a ruminação, em posição deitada e em pé, além do número de passos, número de mudanças de pé para deitado (período deitado) e atividade de movimento total. Todos os dados comportamentais foram reduzidos para -10 dias antes da data do parto para análise estatística usando o procedimento de modelo misto e ANOVA. Os resultados mostraram que o tempo de alimentação e de ruminação diminuiu significativamente (P<0,05) de -10 dias para -1 dia antes do parto, indicando a previsão de parto. Além disso, o tempo de ruminação apresentou seu menor valor (P<0,001) 2 horas antes do parto, e tais mudanças comportamentais podem ser úteis para predizer o parto em búfalas. Da mesma forma, o período deitado e o tempo em pé diminuíram abruptamente (P<0,05) de -3 dias para -1 dia antes do parto, enquanto o tempo deitado aumentou abruptamente (P<0,01) de -3 dias para -1 dia antes do parto (531,57 ± 23,65 para 665,62± 18,14, respectivamente). O número de passos dados e o movimento total aumentaram significativamente (P<0,05) de -10 para -1 dias antes do parto. O tempo de alimentação foi significativamente (P<0,02) reduzido em búfalas de 3ª paridade em comparação com búfalas de 2ª, 4ª e 5ª paridade, enquanto o tempo de espera de búfalas de 5ª paridade foi reduzido (P<0,05) em comparação com búfalas de 2ª a 4ª paridade em -1 dia antes do parto. No entanto, ruminação, posição deitada, número de passos dados e atividade de movimento total em -1 dia antes do parto foram independentes (P>0,05) da paridade em búfalas. A análise de rede neural para variáveis ​​combinadas da tecnologia NEDAP no nível diário produziu 100% de sensibilidade e 98% de especificidade. Em conclusão, a tecnologia NEDAP pode ser usada para medir mudanças comportamentais -10 dias antes do parto, pois pode servir como um guia útil para prever a data do parto em búfalas.(AU)

The neuroprotective effects of Lutongkeli in traumatic brain injury rats by anti-apoptosis mechanism

Purpose: To explore the neuroprotective effects of Lutongkeli (LTKL) in traumatic brain injury (TBI) and detect the related mechanism. Methods: TBI model was established with LTKL administration (2 and 4 g/kg/d, p.o.). Motor function of rats was examined by Rotarod test. Nissl staining was used to show neuron morphology. Furthermore, the disease-medicine common targets were obtained with the network pharmacology and analyzed with Kyoto Encyclopedia of Genes and Genomes. Lastly, the predicted targets were validated by real-time polymerase chain reaction. Results: After LTKL administration, neural behavior was significantly improved, and the number of spared neurons in brain was largely increased. Moreover, 68 bioactive compounds were identified, corresponding to 148 LTKL targets; 2,855 genes were closely associated with TBI, of which 87 overlapped with the LTKL targets and were considered to be therapeutically relevant. Functional enrichment analysis suggested LTKL exerted its pharmacological effects in TBI by modulating multiple pathways including apoptosis, inflammation, etc. Lastly, we found LTKL administration could increase the mRNA level of Bcl-2 and decrease the expression of Bax and caspase-3. Conclusions: This study reported the neuroprotective effect of LTKL against TBI is accompanied with anti-apoptosis mechanism, which provides a scientific explanation for the clinical application of LTKL in the treatment of TBI.

Computational classification of animals for a highway detection system / Classificação computacional de animais para sistema de detecção nas rodovias

Vehicle-animal collisions represent a serious problem in roadway infrastructure. To avoid these roadway collisions, different mitigation systems have been applied in various regions of the world. In this article, a system for detecting animals on highways is presented using computer vision and machine learning algorithms. The models were trained to classify two groups of animals: capybaras and donkeys. Two variants of the convolutional neural network called Yolo (You only look once) were used, Yolov4 and Yolov4-tiny (a lighter version of the network). The training was carried out using pre-trained models. Detection tests were performed on 147 images. The accuracy results obtained were 84.87% and 79.87% for Yolov4 and Yolov4-tiny, respectively. The proposed system has the potential to improve road safety by reducing or preventing accidents with animals.(AU)

As colisões entre veículos e animais representam um sério problema na infraestrutura rodoviária. Para evitar tais acidentes, medidas mitigatórias têm sido aplicadas em diferentes regiões do mundo. Neste projeto é apresentado um sistema de detecção de animais em rodovias utilizando visão computacional e algoritmo de aprendizado de máquina. Os modelos foram treinados para classificar dois grupos de animais: capivaras e equídeos. Foram utilizadas duas variantes da rede neural convolucional chamada Yolo (você só vê uma vez) ­ Yolov4 e Yolov4-tiny (versão mais leve da rede) ­ e o treinamento foi realizado a partir de modelos pré-treinados. Testes de detecção foram realizados em 147 imagens e os resultados de precisão obtidos foram de 84,87% e 79,87% para Yolov4 e Yolov4-tiny, respectivamente. O sistema proposto tem o potencial de melhorar a segurança rodoviária reduzindo ou prevenindo acidentes com animais.(AU)

Computational classification of animals for a highway detection system / Classificação computacional de animais para sistema de detecção nas rodovias

Vehicle-animal collisions represent a serious problem in roadway infrastructure. To avoid these roadway collisions, different mitigation systems have been applied in various regions of the world. In this article, a system for detecting animals on highways is presented using computer vision and machine learning algorithms. The models were trained to classify two groups of animals: capybaras and donkeys. Two variants of the convolutional neural network called Yolo (You only look once) were used, Yolov4 and Yolov4-tiny (a lighter version of the network). The training was carried out using pre-trained models. Detection tests were performed on 147 images. The accuracy results obtained were 84.87% and 79.87% for Yolov4 and Yolov4-tiny, respectively. The proposed system has the potential to improve road safety by reducing or preventing accidents with animals.(AU)

As colisões entre veículos e animais representam um sério problema na infraestrutura rodoviária. Para evitar tais acidentes, medidas mitigatórias têm sido aplicadas em diferentes regiões do mundo. Neste projeto é apresentado um sistema de detecção de animais em rodovias utilizando visão computacional e algoritmo de aprendizado de máquina. Os modelos foram treinados para classificar dois grupos de animais: capivaras e equídeos. Foram utilizadas duas variantes da rede neural convolucional chamada Yolo (você só vê uma vez) ­ Yolov4 e Yolov4-tiny (versão mais leve da rede) ­ e o treinamento foi realizado a partir de modelos pré-treinados. Testes de detecção foram realizados em 147 imagens e os resultados de precisão obtidos foram de 84,87% e 79,87% para Yolov4 e Yolov4-tiny, respectivamente. O sistema proposto tem o potencial de melhorar a segurança rodoviária reduzindo ou prevenindo acidentes com animais.(AU)

Multivariate modeling strategies to predict nutritional requirements of essential amino acids in semiheavy second-cycle hens

An experiment with 23 diets was performed to evaluate the effect of digestible lysine (Lys), digestible methionine + cysteine (Met+Cys), and digestible threonine (Thr) on egg production of H&N Brown second-cycle laying hens (SCLH) for 20 weeks (92-111 weeks of age) in cages under environmental conditions. Body weight (BW), feed intake (FI), feed conversion ratio (FCR), egg weight (EW), number of hen-housed eggs, and livability were also evaluated during the experiment. Diets were formulated from a central composite design that combined five levels of Lys, Met+Cys, and Thr ranging from 727 to 1159, 662 to 1055, and 552 to 882 mg/kg, respectively. Egg production (EP) data were evaluated through three different modeling strategies: egg production models, multivariate polynomial models, and artificial neural networks (ANN). A cascade-forward neural network with logsigmoid transfer function was selected as the best model according to goodness-offit statistics in both identification and validation data. One of the best scenarios for EP of H&N Brown SCLH under specific outdoor conditions was established at Lys, Met+Cys, and Thr levels of 1138, 1031, and 717 mg/hen·day, respectively. The ANN model may be an appropriate tool to study and predict EP of H&N Brown SCLH based on the combination of three different levels of essential digestible amino acids. The strategies included in this work may contribute to improving poultry performance based on modeling techniques to study other production parameters in terms of different nutritional requirements and productive conditions.

Milk fraud by the addition of whey using an artificial neural network / Identificação de fraude de leite por adição de soro de leite usando redes neurais artificiais

The adulteration of milk by the addition of whey is a problem that concerns national and international authorities. The objective of this research was to quantify the whey content in adulterated milk samples using artificial neural networks, employing routine analyses of dairy milk samples. The analyses were performed with different concentrations of whey (0, 5, 10, and 20%), and samples were analyzed for fat, non-fat solids, density, protein, lactose, minerals, and freezing point, totaling 164 assays, of which 60% were used for network training, 20% for network validation, and 20% for neural network testing. The Garson method was used to determine the importance of the variables. The neural network technique for the determination of milk fraud by the addition of whey proved to be efficient. Among the variables of highest relevance were fat content and density.(AU)

A adulteração do leite pela adição de soro de leite é um problema que diz respeito às autoridades nacionais e internacionais. O objetivo deste trabalho foi quantificar o teor de soro em amostras de leite adulterado por meio de redes neurais artificiais, usando como variáveis de entrada os resultados de análises rotineiras em amostras de leite. As análises foram realizadas com diferentes concentrações em relação à adição de soro de leite (0, 5, 10 e 20%), e as amostras foram analisadas quanto à gordura, sólidos não gordurosos, densidade, proteína, lactose, minerais e ponto de congelamento, totalizando 164 ensaios, dos quais 60% foram utilizados para treinamento em rede, 20% para validação de rede e 20% para teste de rede neural. O método de Garson foi utilizado para determinar a importância das variáveis. A técnica de redes neurais para a determinação da fraude ao leite por adição de soro provou ser eficiente. Entre as variáveis de maior relevância estavam o teor de gordura e a densidade.(AU)

Chlorophyll content for millet leaf using hyperspectral imaging and an attention-convolutional neural network / Estimativa do teor de clorofila para folhas de milheto usando imagens hiperespectrais e uma rede neural atenção-convolucional

Chlorophyll is a major factor affecting photosynthesis; and consequently, crop growth and yield. In this study, we devised a chlorophyll-content detection model for millet leaves in different stages of growth based on hyperspectral data. The hyperspectral images of millet leaves were obtained under a wavelength range of 380-1000 nm using a hyperspectral imager. Threshold segmentation was performed with near-infrared (NIR) reflectance and normalized difference vegetation index (NDVI) to intelligently acquire the regions of interest (ROI). Furthermore, raw spectral data were preprocessed using multivariate scatter correction (MSC). A correlation coefficient-successive projections algorithm (CC-SPA) was used to extract the characteristic wavelengths, and the characteristic parameters were extracted based on the spectral and image information. A partial least squares regression (PLSR) prediction model was established based on the single characteristic parameter and multi-characteristic parameter fusion. The determination coefficient (Rv 2) and the root-mean-square error (RMSEv) of the validation set for the multi-characteristic parameter fusion model were reported to be 0.813 and 1.766, respectively, which are higher than those obtained by the single characteristic parameter model. Based on the multi-characteristic parameter fusion, an attention-convolutional neural network (attention-CNN) (Rv 2 = 0.839, RMSEv = 1.451, RPD = 2.355) was established, which is more effective than the PLSR (Rv 2 = 0.813, RMSEv = 1.766, RPD = 2.167) and least squares support vector machine (LS-SVM) models (Rv 2 = 0.806, RMSEv = 1.576, RPD = 2.061). These results indicated that the combination of hyperspectral imaging and attention-CNN is beneficial to the application of nutrient element monitoring of crops.(AU)

A clorofila é um fator importante que afeta a fotossíntese e, consequentemente, o crescimento e o rendimento das culturas. Neste estudo, um modelo de detecção de conteúdo de clorofila é construído para folhas de milheto em diferentes estágios de crescimento, com base em dados hiperespectrais. As imagens hiperespectrais dos diferentes estágios de crescimento das folhas de milheto foram obtidas para 380-1000 nm, utilizando um gerador de imagens hiperespectrais. Uma segmentação de limiar foi realizada com refletância no infravermelho próximo (NIR) e índice de vegetação com diferença normalizada (NDVI) para adquirir de forma inteligente as regiões de interesse (ROI). Além disso, os dados espectrais brutos foram pré-processados usando o método de correção de dispersão multivariada (MSC). Um algoritmo de projeção de coeficiente de correlação sucessivo (CC-SPA) foi utilizado para extrair os comprimentos de onda característicos, e os parâmetros característicos foram extraídos com base nas informações espectrais e de imagem. O modelo de previsão de regressão parcial dos mínimos quadrados (PLSR) foi estabelecido com base nos parâmetros de característica única e na fusão de parâmetros de característica múltipla. O coeficiente de determinação (Rv2) e o erro quadrático médio da raiz (RMSEv) do conjunto de validação para o modelo de fusão de parâmetros com várias características foram obtidos como 0,813 e 1,766, sendo melhores do que os do modelo de parâmetro de característica única. Com base na fusão de parâmetros com várias características, foi estabelecida uma rede neural atenção-convolucional (atenção-CNN) (Rv2 = 0,839, RMSEv = 1,451, RPD = 2,355) mais eficaz que o PLSR (Rv2 = 0,813, RMSEv = 1,766, RPD = 2,167) e mínimos quadrados que suportam modelos de máquina de vetores (LS-SVM) (Rv2 = 0,806, RMSEv = 1,576, RPD = 2,061)....(AU)

Residual nets for understanding animal behavior

Analysis of animal behavior requires proper algorithms for the extraction of desired information from videos. Animal behavior involves various states like facial expression, body movement etc. With the advancement in hardware, deep learning has become popular for analyzing the complex and large dataset. Deep learning algorithms have proved their significance on the benchmark dataset. In this paper, we used Residual Nets for classifying three-hour video containing egg laying induced activity changes in Drosophila. We obtained 99.5% accuracy and found significant improvement in accuracy as compared to CNN (Convolutional Neural Networks). Further, it is suggested that this technique can be used for analysis of animal behavior as well as activities of other domain like object detection, speech recognition, and character recognition, among others.(AU)

Residual nets for understanding animal behavior

Analysis of animal behavior requires proper algorithms for the extraction of desired information from videos. Animal behavior involves various states like facial expression, body movement etc. With the advancement in hardware, deep learning has become popular for analyzing the complex and large dataset. Deep learning algorithms have proved their significance on the benchmark dataset. In this paper, we used Residual Nets for classifying three-hour video containing egg laying induced activity changes in Drosophila. We obtained 99.5% accuracy and found significant improvement in accuracy as compared to CNN (Convolutional Neural Networks). Further, it is suggested that this technique can be used for analysis of animal behavior as well as activities of other domain like object detection, speech recognition, and character recognition, among others.

Redes neurais artificiais na estimação de variáveis biométricas de mudas de espécies florestais produzidas em diferentes substratos / Artificial neural networks for estimating the biometric variables of seedlings from forest species produced in different substrates

O objetivo do trabalho foi avaliar o crescimento em diâmetro do coleto e altura, e a produção de matéria seca total de mudas de Myracrodruon urundeuva, Jacaranda brasiliana e Mimosa caesalpiniaefolia. Concomitantemente, desenvolveu-se uma Rede Neural Artificial (RNA) do tipo Multilayer Perceptron que seria capaz de estimar a H e a MST das mudas das espécies estudadas. As mudas foram cultivadas em ambiente protegido com 50% de sombra. Assim, os tratamentos foram considerados com cinco proporções do material orgânico (0, 20, 40, 60 e 80% v/v) na composição do substrato final (solo da área desertificada). Aos 120 dias após a semeadura, as mudas foram coletadas para determinação das variáveis biométricas. A rede MLP foi utilizada empregando-se o algoritmo de treinamento Levenberg-Marquardat. As variáveis utilizadas como entrada da MLP para a estimação da altura e massa seca das mudas foram: diâmetro do coleto, diâmetro mínimo, médio e máximo do coleto, as espécies e fontes de resíduos orgânicos (esterco bovino, esterco caprino e palha de arroz), totalizando dez entradas. Foi utilizada a função de ativação tangente hiperbólica. Como resultados, recomenda-se a proporção 80:20% (esterco bovino e/ou esterco caprino:solo da área degradada) ao substrato de cultivo para o crescimento das mudas das espécies. A adição de doses de esterco bovino e esterco caprino influenciaram o DC do...

The aim of this study was to evaluate the stem growth in diameter and height as well as the production of total dry matter from seedlings of Myracrodruon urundeuva, Jacaranda brasiliana and Mimosa caesalpiniaefolia. Concurrently, an Artificial Neural Network (RNA) of Multilayer Perceptron type that would be able to estimate the H and the MST of the seedlings of the studied species was developed. The seedlings were cultivated in a protected environment with 50% shade. Thus, the treatments were considered with five proportions of the organic material (0, 20, 40, 60 and 80% v/v) in the final substrate composition (desertified area soil). At 120 days after sowing, the seedlings were collected to determine the biometric variables. The MLP network was used with help of the Levenberg-Marquardat training algorithm. The variables used as input of the MLP for height and dry mass estimation of the seedlings were: stem diameter, minimum, medium and maximum diameter of stem; and species and sources of organic residues (cattle manure, goat manure and rice straw), totaling ten entries. The hyperbolic tangent activation function was conducted. As a result, a 80:20% ratio (bovine manure and/or goat manure: soil from the degraded area) is recommended to be used in the growing substrate for seedling growth. The addition of bovine manure and goat manure doses influenced the Jacaranda brasiliana DC...

Redes neurais artificiais na estimação de variáveis biométricas de mudas de espécies florestais produzidas em diferentes substratos / Artificial neural networks for estimating the biometric variables of seedlings from forest species produced in different substrates

O objetivo do trabalho foi avaliar o crescimento em diâmetro do coleto e altura, e a produção de matéria seca total de mudas de Myracrodruon urundeuva, Jacaranda brasiliana e Mimosa caesalpiniaefolia. Concomitantemente, desenvolveu-se uma Rede Neural Artificial (RNA) do tipo Multilayer Perceptron que seria capaz de estimar a H e a MST das mudas das espécies estudadas. As mudas foram cultivadas em ambiente protegido com 50% de sombra. Assim, os tratamentos foram considerados com cinco proporções do material orgânico (0, 20, 40, 60 e 80% v/v) na composição do substrato final (solo da área desertificada). Aos 120 dias após a semeadura, as mudas foram coletadas para determinação das variáveis biométricas. A rede MLP foi utilizada empregando-se o algoritmo de treinamento Levenberg-Marquardat. As variáveis utilizadas como entrada da MLP para a estimação da altura e massa seca das mudas foram: diâmetro do coleto, diâmetro mínimo, médio e máximo do coleto, as espécies e fontes de resíduos orgânicos (esterco bovino, esterco caprino e palha de arroz), totalizando dez entradas. Foi utilizada a função de ativação tangente hiperbólica. Como resultados, recomenda-se a proporção 80:20% (esterco bovino e/ou esterco caprino:solo da área degradada) ao substrato de cultivo para o crescimento das mudas das espécies. A adição de doses de esterco bovino e esterco caprino influenciaram o DC do...(AU)

The aim of this study was to evaluate the stem growth in diameter and height as well as the production of total dry matter from seedlings of Myracrodruon urundeuva, Jacaranda brasiliana and Mimosa caesalpiniaefolia. Concurrently, an Artificial Neural Network (RNA) of Multilayer Perceptron type that would be able to estimate the H and the MST of the seedlings of the studied species was developed. The seedlings were cultivated in a protected environment with 50% shade. Thus, the treatments were considered with five proportions of the organic material (0, 20, 40, 60 and 80% v/v) in the final substrate composition (desertified area soil). At 120 days after sowing, the seedlings were collected to determine the biometric variables. The MLP network was used with help of the Levenberg-Marquardat training algorithm. The variables used as input of the MLP for height and dry mass estimation of the seedlings were: stem diameter, minimum, medium and maximum diameter of stem; and species and sources of organic residues (cattle manure, goat manure and rice straw), totaling ten entries. The hyperbolic tangent activation function was conducted. As a result, a 80:20% ratio (bovine manure and/or goat manure: soil from the degraded area) is recommended to be used in the growing substrate for seedling growth. The addition of bovine manure and goat manure doses influenced the Jacaranda brasiliana DC...(AU)

Rede Neural Artificial para diferenciação entre soro de muçarela de búfala e vaca por meio de Espectroscopia no Infravermelho com Transformada de Fourier / Artificial Neural Network for differentiation between buffalo and cow mozzarella whey by Infrared Spectroscopy with Fourier Transform

Objetivou-se diferenciar amostras de soro obtidas de muçarela de leites de vaca, búfala e com misturas de leites entre as espécies. Foram elaboradas formulações com leite de búfala, vaca e com inclusões crescentes de leite bovino ao bubalino, às quais foram utilizadas na produção de queijo muçarela, gerando o soro a ser analisado. As amostras de soro foram avaliadas a partir da Espectroscopia no Infravermelho com Transformada de Fourier, associada a Análise de Componentes Principais e Redes Neurais Artificiais. Foi verificada a separação espacial do soro dos tratamentos búfala e vaca. A Rede Neural garantiu a identificação da presença de leite de vaca nas amostras de soro avaliadas, porém sem indicar com precisão os níveis de adulteração.

A smartphone-based apple yield estimation application using imaging features and the ANN method in mature period

Apple yield estimation using a smartphone with image processing technology offers advantages such as low cost, quick access and simple operation. This article proposes distribution framework consisting of the acquisition of fruit tree images, yield prediction in smarphone client, data processing and model calculation in server client for estimating the potential fruit yield. An image processing method was designed including the core steps of image segmentation with R/B value combined with V value and circle-fitting using curvature analysis. This method enabled four parameters to be obtained, namely, total identified pixel area (TP), fitting circle amount (FC), average radius of the fitting circle (RC) and small polygon pixel area (SP). A individual tree yield estimation model on an ANN (Artificial Neural Network) was developed with three layers, four input parameters, 14 hidden neurons, and one output parameter. The system was used on an experimental Fuji apple (Malus domestica Borkh. cv. Red Fuji) orchard. Twenty-six tree samples were selected from a total of 80 trees according to the multiples of the number three for the establishment model, whereby 21 groups of data were trained and 5 groups o data were validated. The R2 value for the training datasets was 0.996 and the relative root mean squared error (RRMSE) value 0.063. The RRMSE value for the validation dataset was 0.284 Furthermore, a yield map with 80 apple trees was generated, and the space distribution o the yield was identified. It provided appreciable decision support for site-specific management.(AU)