We consider gaining from information of variable quality that might be gotten from various heterogeneous sources. Because of rich semantics of the information and absence of from the earlier learning about the investigation undertaking, intemperate disinfection is frequently important to guarantee security, prompting huge loss of the information utility. Security Preserving Data Mining (PPDM) helps to mine data and uncovers designs from extensive dataset shielding private and touchy information from being uncovered. With the approach of shifted advancements in information gathering, stockpiling and preparing, various security protection strategies have been created. We propose a suitable security shielding K-infers gathering plan that can be viably outsourced to cloud servers. The present work grants cloud servers to perform packing particularly completed encoded datasets, while accomplishing similar computational many-sided quality and precision contrasted and grouping over decoded ones. Guide Reduce approach likewise consolidated in this paper, which makes this work extraordinarily fitting for Map Reduce condition. Differentially security approach guarantees the aftereffects of inquiries to a database, which will grow the flexibility and time capability over existing techniques. Interestingly of elective arrangements, dp- GAN features an arrangement of key highlights. It gives hypothetical security ensure by means of upholding the differential protection standard. It holds attractive utility in the discharged model, empowering an assortment of generally unthinkable investigations; and above all, it accomplishes reasonable preparing adaptability and steadiness by utilizing multifold streamlining techniques. We propose a technique for changing the learning rate as a component of the heterogeneity, and demonstrate new lament limits for our strategy in two instances of premium. At long last, we assess the execution of our calculation on genuine information.