Ciphertext attribute-based encryption is a proven mechanism for providing the privacy and security for the shared resources in the cloud. However, the issues that are concerned with the sharing mechanisms such as master key and access policies were exploited by the malicious users. Moreover, the access control mechanisms are developed by using the large universe of attributes of the shared resource in the cloud. More number of attributes results into increase in computation time while computing the master and secret keys as well as for encryption and decryption processes. The observations over the participating attributes play vital role to prepare a machine learning model in terms of better accountability. In this paper we have proposed specific attribute-based encryption to provide the better security and better cloud access control mechanism. Inclusion of dynamic attributes while performing the encryption at data owner, cloud server would serve a better performance to avoid key exposure. This performance is elevated while generating the secret key at the proxy server. The performance has been found to be satisfactorily encouraging by reducing the computation time to almost half of the existing schemes and the observations are in accordance to the required accountability.