Tracking the Connection Between Brazilian Agricultural Diversity and Native Vegetation Change by a Machine Learning Approach

Silva, Marcos Aurélio Santos da; Matos, Leonardo Nogueira; Santos, Flavio Emanuel de Oliveira; Dompieri, M. H. G.; Moura, Fábio Rodrigues de

doi:10.1109/tla.2022.9904762

Cited by 7 publications

(10 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Their findings showed evidence that lowly diversified regions tend to present a low degree of sustainability, also observed by [Fatch et al 2021]. [Silva et al 2022] used the IBGE's estimates of annual agricultural production from 1999 to 2018 and calculated a diversity index based on Shannon's entropy for each category (animal herd, planted area with temporary crops, the production value for temporary and permanent crops, aquaculture, silviculture, vegetal extractivism, and animal), so eight variables for 20 years and 5570 municipalities. Then, they applied a shallow learning algorithm based on the Self-Organizing Map Artificial Neural Network combined with the k-means to cluster the spatial panel data.…”

Section: Introductionmentioning

confidence: 56%

“…Designing territorial public policies demands the identification of regional particularities, and we can achieve this by clustering the regions according to their agricultural production similarities. [Silva et al 2022] combined featuring engineering and clustering analysis to divide the Brazilian municipalities into eight agricultural production diversity trends groups, linking each group to a different level of native vegetation change. Their findings showed evidence that lowly diversified regions tend to present a low degree of sustainability, also observed by [Fatch et al 2021].…”

Section: Introductionmentioning

confidence: 99%

“…The raw spatial panel data used by [Silva et al 2022] comprises 196 variables for 20 years. The use of a feature engineering method reduced from 196 to 8 variables and diminished some data challenges as a massive presence of zeros and lots of values near zero.…”

Section: Introductionmentioning

confidence: 99%

“…So, the research questions are: What if we performed a clustering analysis directly over the raw panel data? How could a Deep Learning (DL) feature extraction substitute the feature engineering used by [Silva et al 2022]?…”

Section: Introductionmentioning

confidence: 99%

“…This paper applied a Deep Learning strategy based on autoencoders to feature extract on the raw Brazilian agricultural spatial panel data to cluster the data, considering the clustering obtained by [Silva et al 2022] as the ground truth. We used two classical clustering algorithms over the extracted features, k-means and Self-Organizing Maps associated with the k-means.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Feature engineering vs. extraction: clustering Brazilian municipalities through spatial panel agricultural data via autoencoders

Silva¹,

Matos²,

Santos³

et al. 2022

Anais Do XIX Encontro Nacional De Inteligência Artificial E Computacional (ENIAC 2022)

View full text Add to dashboard Cite

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.

show abstract

Section: Introductionmentioning

confidence: 56%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations