With the emergence of cloud computing, a large amount of private data is stored and processed in the cloud. However, Cloud computing is often confronted with several threats such as the insecure interface, abuse of cloud computing, data loss, anonymous database hacking, crashing, and malicious insiders. Existing security problems in the cloud, such as the security for the cloud framework, location privacy in the mobile cloud, security in cloud storage, data mining, and machine learning.
   Moreover, Traditional privacy-preserving techniques are based mainly on data perturbation and cryptographic methods. Despite that, Data users or owners upload encrypted data to the cloud or third-party platforms, such as Google Cloud, Amazon, and so on. Conventionally, data must be decrypted before analyzing, which raises privacy concerns. However, due to the energy-constrained devices, decryption of some big data in the cloud often requires enormous computation resources. Thus, Data privacy has to be preserved if encrypted data are directly classified on the cloud or IoT devices without decryption. To resolve the privacy-preserving issue, the deep learning model has outsourced encrypted data owned by multiple users in the cloud and leveraged the computing ability as the computational cost and the communication cost of the data owners effectively.