Spectral clustering is a useful method in artificial intelligence, but it can be slow and resource-intensive. For those with limited data and computational power, using cloud computing to handle the heavy lifting can be very helpful. However, sending data to the cloud raises privacy concerns. In this project, we tackle this problem by developing a secure way to use spectral clustering in a multi-user environment. We use a special encryption method called CKKS to keep data private while it's being processed. Our approach ensures that users only need to upload encrypted data and do not interact with each other or the cloud server directly. We also show that our method is both accurate and efficient through theoretical analysis and experiments.

Keywords: CKKS Encryption and decryption, cloud computing, homomorphic encryption, privacy, spectral clustering.