An optimized deep neural network-based financial statement fraud detection in text mining

Identifying Financial Statement Fraud (FSF) events is very crucial in text mining. The researcher’s community mostly utilized the data mining method for detecting FSF. In this direction, quantitative has by research i.e. financial ratio presented fraud financial statements. On investigation there no researches like auditor's remarks present published reports. For reason, paper develops optimized deep neural network-based FSF detection qualitative pre-processing of performed initially using filtering, lemmatization, and tokenization. Then, feature selection done Harris Hawks Optimization (HHO) algorithm. Finally, a Deep Neural Network-Based Deer Hunting Optimization (DNN-DHO) to identify or no-fraud report The developed detection methodology executed Python environment statement datasets. output approach gives high classification accuracy (96%) comparison to standard classifiers DNN, CART, LR, SVM, Bayes, BP-NN, KNN. Also, it provides better outcomes all performance metrics.