Marcos Aurélio Santos da Silva

Este trabalho avalia um método de seleção de Mapas Auto-Organizados a partir da qualidade da preservação topológica e representação dos dados de entrada da rede neural para fins de clusterização de dados de benchmark a partir da segmentação dos pesos da rede neural. Foram avaliados cinco algoritmos de clusterização: k-médias, hierárquico aglomerativo e três métodos baseados no particionamento de grafos. Os resultados mostraram que o método de seleção da melhor rede neural foi efetivo para todas as quatro bases de dados avaliadas, embora não tenha gerado resultados ótimos. Observou-se que o desempenho dos algoritmos de clusterização varia conforme o tipo de dado, com o k-médias apresentando bom desempenho para dados hiperesféricos e para a base Iris, o método hierárquico aglomerativo sendo mais efetivo para a base MNIST e um método baseado no particionamento de grafos mais efetivo para dados com estrutura arbitrária.

The objective of this work was to evaluate the impact of extreme and exceptional droughts, classified by Agência Nacional de Águas e Saneamento Básico (ANA), on agricultural production diversity, measured by the Shannon index, from 2014 to 2023. Data from the drought monitor of ANA and from Instituto Brasileiro de Geografia e Estatística were used to calculate diversity indices at the municipal level in Brazil. The unconditional quantile regression was applied to evaluate the effects of droughts on diversified systems, considering the production value of temporary and permanent crops, as well as livestock inventory. The obtained results reveal that less diversified municipalities are more vulnerable, particularly in the Northeastern region. Permanent crops show a higher resilience, while temporary crops sustain losses even in diversified systems. Livestock exhibits adaptive responses, with an increased diversity in some cases. Although diversification reduces vulnerabilities, it requires complementary policies for annual crops and targeted support to the Northeast.

This study aimed to manipulate databases from the Landsat-8 project to select orbital scenes with minimal cloud coverage aligned with the growth periods of cotton, corn, and soybean crops in the Cerrado biome. Employing an Extract, Transform, Load (ETL) approach, the research focused on extracting and transforming large datasets by integrating georeferenced information from IBGE and data from the Agricultural Risk Zoning (ZARC). The planting windows for the 2014 to 2019 harvests were identified with a 20% or lower climate risk, considering variables such as soil and climate conditions. The results indicated that the availability of images with low cloud coverage is inversely related to the planting windows, especially during critical periods. To streamline the processing, algorithms were developed in R and Python, and a script for automatic image downloading was implemented. Despite data standardization challenges, applying big data techniques proved essential for the analysis. This study enhances efficiency in agricultural planning and emphasizes the importance of further exploring remote sensing technologies in response to current food security and sustainability demands.

Este artigo propõe um algoritmo para análise de agrupamentos de mapas temáticos com classes categóricas ordinais para suporte ao zoneamento territorial. O método proposto combina o Mapa Auto-Organizável com técnicas de segmentação de grafos para clusterização dos dados. A abordagem foi avaliada com dados sintéticos e aplicada no zoneamento ambiental da bacia do Alto Taquari, MS/MT. Os resultados mostraram a capacidade do algoritmo separar os dados em grupos diferenciáveis unimodais.

Brazilian agricultural production presents high spatial diversity, challenging the conception of public policies. This article proposes an approach for grouping Brazilian municipalities according to their agricultural production. We combine a feature extraction using autoencoders and clustering based on k-means and Self-Organizing Maps. We used panel data from IBGE’s annual estimates of the production value of permanent and temporary crops, animal products, aquaculture, plant extractivism, forestry, planted areas, and herd population between 1999 and 2018. We analyzed different structures of simple stacked and incomplete autoencoders, varying the number of layers and neurons in each, and evaluated the asymmetric exponential linear loss function to handle the sparse data. We applied the Isomap, Kernel PCA, Truncated SVD, and MDS dimensionality reduction methods for comparative purposes. Results showed that the autoencoders could extract characteristics from the transformed raw data to allow the clustering of municipalities to reveal regional and evenintra-regional patterns. The autoencoders improved comparative performance as the intrinsic dimensionality increased.

Brazilian agricultural production presents a high degree of spatial diversity, which challenges designing territorial public policies to promote sustainable development. This article proposes a new approach to cluster Brazilian municipalities according to their agricultural production. It combines a feature extraction mechanism using Deep Learning based on Autoencoders and clustering based on k-means and Self-Organizing Maps. We clustered the panel data from IBGE’s annual estimates of Brazilian agricultural production between 1999 and 2018. The results show that in comparison with the ground truth adopted, the autoencoder model combined with the Self-Organizing Maps and the k-means algorithm presented a better result than clustering the raw data using k-means. It demonstrated the ability of simple stacked autoencoders to reduce the dimensionality and create a new space of features in their latent layer where the data can be analyzed and clustered.

This article compares the clustering of Brazilian municipalities according to their agricultural diversity using two approaches, one based on feature engineering and the other based on feature extraction using Deep Learning based on autoencoders and cluster analysis based on k-means and Self-Organizing Maps. The analyzes were conducted from panel data referring to IBGE’s annual estimates of Brazilian agricultural production between 1999 and 2018. Different structures of simple stacked undercomplete autoencoders were analyzed, varying the number of layers and neurons in each of them, including the latent layer. The asymmetric exponential linear loss function was also evaluated to cope with the sparse data. The results show that in comparison with the ground truth adopted, the autoencoder model combined with the k-means presented a superior result than the clustering of the raw data from the k-means, demonstrating the ability of simple autoencoders to represent from their latent layer important features of the data. Although the general accuracy is low, the results are promising, considering that we evaluated the most simple strategy for Deep Clustering.

In Brazil, agribusiness has a considerable role in the country's GDP. Because of this, the State needs territorial planning to minimize the impacts on natural resources, especially in the Pantanal and Amazon biomes, where agribusiness has expanded. The lower the agricultural diversification, the lower the pattern of land use homogeneity, generally associated with agribusiness, especially when it occupies large areas with more technological productive units. This paper investigates the relationship between spatial diversification patterns and the dynamics of native vegetation in Brazil. We propose a feature engineering and clustering approach for 5570 Brazilian municipalities between 1999 and 2018. It was based on the unsupervised artificial neural network Self-Organizing Map (SOM) to divide the municipalities into homogeneous groups of agricultural products diversity trends. The results were compared with the change in vegetation area using data from the national land use-mapping project called Mapbiomas. The analysis allowed the identification of three different regimes of modification in native vegetation, particularly related to municipalities in Brazil's Midwest and North regions, indicating substantial changes in the Cerrado and Amazon biomes.

Empowerment of civil society, beyond its own value, seems to be a nice lever for the efficiency of sustainable development programs and the Brazilian government has embarked on this path. If economic results are quite easy to measure, this is not the case for the evolution of power relationships toward a better balance between social actors' positions. This article shows how to apply the SocLab framework, a modeling approach that considers power relationships within systems of organized action, to shed light on this issue in The Southern Rural Territory of Sergipe. The simulation results of the model showed significant changes from 2010 to 2016 mainly in the capability, but also in the power of the Associations despite their internal problems to put into practice their ability of negotiation or even of real engagement in the territory.