UNTRUSTED CLOUD OUTSOURCING FOR LDA-BASED FACE RECOGNITION

In practical applications, face recognition has been widely used in public security and attendance systems. Many clients find itextremely challenging to use the linear discriminant analysis (LDA) method on their resource-constrained devices, such as laptopcomputers and cellphones, despite it being one of the most important algorithms in the field of face recognition. Contracting outcalculations gives clients a viable way to complete demanding tasks with less processing power. In this work, we develop a protocolfor outsourcing LDA-based face recognition to an untrusted cloud that can assist the client in concurrently performing matrixmultiplication (MM), matrix inversion (MI), and eigenvalue decomposition (ED). The suggested outsourcing methodology canconceal the client's sensitive information from the cloud. More significantly, the client can use probability one to confirm whetherthe outsourcing results are accurate, making it impossible for the server to mislead the client. Furthermore, the suggested protocolsignificantly reduces the client's processing complexity, allowing the client to efficiently finish the LDA algorithm. Lastly, we putthe technique into practice and provide a thorough assessment. The experimental findings show that the client achieves significantcomputing savings and that the suggested protocol's face recognition accuracy is nearly the same as that of the original LDAalgorithm