Cloud computing, fulfilling the “Computing as Utility” model, is widely popular for providing on-demand services. However, the primary challenge for cloud providers is the increasing vulnerability to cyber-attacks. Traditional detection methods are insufficient, necessitating advanced approaches. This research suggests a mechanism for detecting cyberattacks using deep learning for Cloud Computing, employing a hybrid CNN and Self-Organizing Maps (SOM) methodology. The two-dimensional representation of input space facilitates the identification of assaults known as distributed denial of service (DDoS), achieving superior results compared to existing techniques in terms of lower detection times (1.2 seconds), and higher accuracy (97%), precision (96%), and recall (97%).