IDENTIFICATION OF ANAMOLY FINANCIAL TRANSACTION USING EXPLAINABLE AI AND FEDERATED LEARNING

Author : Praveen

Abstract :In the domain of monetary misrepresentation location, accomplishing a harmony among straightforwardness and security is basic. Conventional methodologies frequently neglect to give both elevated degrees of exactness and clear, justifiable clarifications for navigation, all while safeguarding delicate information. This paper looks at the capability of Reasonable computer based intelligence (XAI) and United Learning (FL) to conquer these difficulties. XAI offers models that empower monetary establishments to decipher and believe forecasts, which is fundamental in exceptionally managed conditions. Then again, Unified Learning considers the preparation of AI models across dispersed information sources, guaranteeing that delicate monetary information stays secure and doesn’t leave its unique area. This study surveys the presentation of various calculations in misrepresentation discovery, including Profound Brain Organizations (DNN), Repetitive Brain Organizations (RNN), Choice Trees, Irregular Timberlands, Angle Helping, and their application inside the combined learning system. Moreover, the paper investigates streamlining strategies like Stochastic Slope Drop (SGD) to work on model union. By joining the benefits of XAI and FL, this exploration features how monetary establishments can further develop misrepresentation identification while keeping up with straightforwardness, safeguarding security, and guaranteeing consistence with information insurance guidelines

Keywords :In the domain of monetary misrepresentation location, accomplishing a harmony among straightforwardness and security is basic. Conventional methodologies frequently neglect to give both elevated degrees of exactness and clear, justifiable clarifications for navigation, all while safeguarding delicate information. This paper looks at the capability of Reasonable computer based intelligence (XAI) and United Learning (FL) to conquer these difficulties. XAI offers models that empower monetary establishments to decipher and believe forecasts, which is fundamental in exceptionally managed conditions. Then again, Unified Learning considers the preparation of AI models across dispersed information sources, guaranteeing that delicate monetary information stays secure and doesn’t leave its unique area. This study surveys the presentation of various calculations in misrepresentation discovery, including Profound Brain Organizations (DNN), Repetitive Brain Organizations (RNN), Choice Trees, Irregular Timberlands, Angle Helping, and their application inside the combined learning system. Moreover, the paper investigates streamlining strategies like Stochastic Slope Drop (SGD) to work on model union. By joining the benefits of XAI and FL, this exploration features how monetary establishments can further develop misrepresentation identification while keeping up with straightforwardness, safeguarding security, and guaranteeing consistence with information insurance guidelines

Conference Name :International Conference on Computer Science and Engineering (ICCSAE-25)

Conference Place Kollam, India

Conference Date 25th May 2025