Data security is one of the most critical parts of considerable data investigation. Most cloud systems deal with sensitive information, such as personal, corporate, or medical records. Threats to this type of data might affect the growth of the cloud systems that store it. On the other hand, traditional security solutions are ineffective in safeguarding substantial data transfers. To handle the vast generation of data and their security aspects necessitated an efficient privacy-preserving mechanism. Initially, cloud-based big data is clustered with the K-Modes-based Fast Mutation Artificial Bee Colony clustering (K-mode FMABC) technique, and clustered information is balanced with the support of the Hadoop-based map-reduce mechanism. Further, the information is encrypted using an advanced super encryption technique, and the convolution process is accomplished over the method of estimation that encrypt data. The estimation process is completed by the deep neural network (DNN). If the encrypted data are not properly encrypted, the data will be sent through the convolution process again. The proposed framework's experimental findings are compared with the current methodologies; the big data-based privacy-preserving scheme outperforms existing techniques.