Taxpayer compliance classification using C4.5, SVM, KNN, Naive Bayes and MLP

O Uso de Inteligência Artificial no Combate à Evasão Fiscal: Uma Revisão Sistemática da Literatura

“…Em [Jupri and Sarno 2018] são comparados 5 algoritmos de classificação no contexto de conformidade fiscal, são eles: Decision Trees C4.5, Support Vector Machine (SVM), KNN, Naive Bayes e Multilayer Preceptron (MPL). O algoritmo que obteve melhor resultado foi o Decision Trees C4.5, com 98.93% de acurácia, enquanto que o MPL teve 89.26% de acurácia, ficando na última posição do ranking.…”

“…Supervisionado , [Jupri and Sarno 2018], [Xiangyu et al 2018], [Zhu et al 2018], [Wu et al 2019], [Rahimikia et al 2017], [Goumagias et al 2018], [Kleanthous and Chatzis 2019] 8 Não supervisionado [de Roux et al 2018], [Wei et al 2019], [Mehta et al 2019], [Vanhoeyveld et al 2020], [Mathews et al 2018], [Mehta et al 2018] 6…”

“…Além disso, o uso de IA também tem sido frequentemente utilizado para combater a inadimplência, de modo a serem asseguradas a regularidade e a previsibilidade da arrecadação [Jupri andSarno 2018, Rahimikia et al 2017]. Paralelo a isso, as previsões de receita através de regressões lineares também foram apontadas no estudo [Mathews et al 2018].…”

A comparative study of sentiment analysis using SVM and SentiWordNet

“…Then for the split between training data and testing data using K-Fold Cross Validation method. So the data are divided into k fold and then will be executed classification process as much as the k and for the testing data is selected from one of k fold and training data is fold which are not used as the data testing [16]. The selection of data testing per round is selected in sequence starting from the folds 1, see Figure 3 for the illustration.…”

Sparsity aware dynamic gesture recognition using radar sensors with angular diversity

“…For the KNN classifier, the recognition result is decided based on a majority vote of k NN, as described in [26]. For the Naïve Bayes classifier, decision‐making on the Bayes theory and independent input variables are applied, as implemented in [27]. In this experiment, we assume that the prior probabilities of these gestures are the same.…”