Abstract

Data hiding is a modern paradigm that efficaciously hides a class or module's implementation details, revealing just the essential information to the outdoors world. It improves security and encapsulation by proscribing access to sensitive records and procedures, lowering unintended intrusion, and supporting modular design principles. In this research, we aim to develop an innovative machine learning (ML)-driven data hiding strategy for financial cloud security. We propose a new Secured Dynamic Paillier Federated Neural Network (SDPFNN) encryption algorithm to overcome the data hiding problem in distributed ML (DML) included economic cloud packages with embedded structures. Initially, research gathered a dataset that includes financial transaction data from various organizations to train our suggested model. The data cleaning procedure is conducted to pre-process the gathered raw data. Our recommended encryption approach prevents individuals from recognizing encrypted data. We present a technique for mapping the original data into high-dimensional space. We conducted results analysis using different metrics to assess the effectiveness of the suggested model. To determine the efficiency of the suggested strategy, research performed a comparative study with other comparable methodologies. The experimental findings demonstrate that the proposed model outperforms conventional approaches.