“…Table 4 Quantity of papers per knowledge area Knowledge area References Qtd % Financial [5], [6], [7], [8], [9], [10], [11], [12], [13], [14], [15], [16], [17], [18], [19], [20], [3], [21], [22], [23], [24], [25], [26] 23 71.88%…”
Section: Evaluated Resultsmentioning
confidence: 99%
“…• Rao et al [21] suggest the use of knowledge graphs (Know-GNN), in a semisuperivsed form, for dealing with data with noise; • Wang et al [24] explore the possibility of union of supervised and non-supervised information for fraud detection. For that, the authors propose the construction of an attribute network which, in combination with a model of hierarchical attention (SemiGNN), can provide interpretable results; • Alam and Ali [26] use graph techniques to bring interpretability to the algorithm of default detection in financial loans, without losing accuracy.…”
Section: Interpretability Techniquementioning
confidence: 99%
“…Chou [9], Jesus et al [16], Salekshahrezaee, Leevy and Khoshgoftaar [50], Przekop [3]; • The necessity to continue the optimizing process of the model, hyper-parameters or application of feature engineering techniques in the data that feeds it, as presented in Kong et al [51], Sávic et al [22], Son et al [23], Hayashi and Oishi [14], Liu, Zhou and He [25] and Yuan et al [34]; • The necessity to test the proposed methodology with other algorithms, such as in La Gatta et al [30], Kiefer and Pesch [17], Pelckmans [52], Rao et al [21] and Abeyrathna, Granmo and Goodwin [27]; • The possibility to explore more interpretability techniques, for example the works from Kou et al [18], Coma-Puig and Carmona [12] and Yuan et al [34].…”
Section: Interpretability Techniquementioning
confidence: 99%
“…About the datasets, the scenario is presented in Table 6. Cited but not made available [5], [7], [8], [9], [10], [12], [17], [18], [19], [3], [21], [22], [23], [24], [32], [25] 16 50.00%…”
In a technological world, in which data is generated exponentially, financial analysis has gradually become more important to avoid large losses due to fraud. Considering the large volume and the difficulty of human data checking, machine learning technologies have become one of the main tools to solve the problem. However, due to the creation of data protection laws in several countries, in some scenarios the detection of fraud through intelligence algorithms becomes insufficient. Therefore, it is necessary to understand how the algorithm actually labels a transaction as fraudulent or not. In this work, presented as a systematic literature review, we look for answers on how explicable/interpretable fraud detection algorithms have been applied in order to solve the problem of illegal activities in the financial sector. As a result of the mapping of the current state of the art, this work highlights the gaps in the literature and present the scenario of interpretable techniques used for fraud detection comprehension.
“…Table 4 Quantity of papers per knowledge area Knowledge area References Qtd % Financial [5], [6], [7], [8], [9], [10], [11], [12], [13], [14], [15], [16], [17], [18], [19], [20], [3], [21], [22], [23], [24], [25], [26] 23 71.88%…”
Section: Evaluated Resultsmentioning
confidence: 99%
“…• Rao et al [21] suggest the use of knowledge graphs (Know-GNN), in a semisuperivsed form, for dealing with data with noise; • Wang et al [24] explore the possibility of union of supervised and non-supervised information for fraud detection. For that, the authors propose the construction of an attribute network which, in combination with a model of hierarchical attention (SemiGNN), can provide interpretable results; • Alam and Ali [26] use graph techniques to bring interpretability to the algorithm of default detection in financial loans, without losing accuracy.…”
Section: Interpretability Techniquementioning
confidence: 99%
“…Chou [9], Jesus et al [16], Salekshahrezaee, Leevy and Khoshgoftaar [50], Przekop [3]; • The necessity to continue the optimizing process of the model, hyper-parameters or application of feature engineering techniques in the data that feeds it, as presented in Kong et al [51], Sávic et al [22], Son et al [23], Hayashi and Oishi [14], Liu, Zhou and He [25] and Yuan et al [34]; • The necessity to test the proposed methodology with other algorithms, such as in La Gatta et al [30], Kiefer and Pesch [17], Pelckmans [52], Rao et al [21] and Abeyrathna, Granmo and Goodwin [27]; • The possibility to explore more interpretability techniques, for example the works from Kou et al [18], Coma-Puig and Carmona [12] and Yuan et al [34].…”
Section: Interpretability Techniquementioning
confidence: 99%
“…About the datasets, the scenario is presented in Table 6. Cited but not made available [5], [7], [8], [9], [10], [12], [17], [18], [19], [3], [21], [22], [23], [24], [32], [25] 16 50.00%…”
In a technological world, in which data is generated exponentially, financial analysis has gradually become more important to avoid large losses due to fraud. Considering the large volume and the difficulty of human data checking, machine learning technologies have become one of the main tools to solve the problem. However, due to the creation of data protection laws in several countries, in some scenarios the detection of fraud through intelligence algorithms becomes insufficient. Therefore, it is necessary to understand how the algorithm actually labels a transaction as fraudulent or not. In this work, presented as a systematic literature review, we look for answers on how explicable/interpretable fraud detection algorithms have been applied in order to solve the problem of illegal activities in the financial sector. As a result of the mapping of the current state of the art, this work highlights the gaps in the literature and present the scenario of interpretable techniques used for fraud detection comprehension.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.