Multi-label financial statement fraud detection based on long short-term memory and multilayer perceptron hybrid model

Key Points

• Financial statement fraud detection improved using a hybrid model combining long short-term memory and multilayer perceptron.
• The analysis achieved an accuracy rate of 92%, highlighting effective fraud identification techniques.
• The hybrid model utilizes multi-label classification to manage various types of fraudulent activities in financial data.
• This advancement supports more accurate fraud detection systems in financial industries, potentially reducing risks.