TabNet is reported to beat DNN, LSTM, GRU and CNN1D on a Kaggle bank-fraud dataset, but the evaluation leaks synthetic SMOTE data across folds and reports in-sample metrics.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
q-fin.GN 1years
2026 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Prediction of bank transaction fraud using TabNet an adaptive deep learning architecture
TabNet is reported to beat DNN, LSTM, GRU and CNN1D on a Kaggle bank-fraud dataset, but the evaluation leaks synthetic SMOTE data across folds and reports in-sample metrics.