Due to Internet facilities, digital information is frequently shared and stored on cloud devices, enabling cyber-attacks. Several organizations are securing their data from cyber threats using recent technologies. However, zero-day intrusions are challenging to detect due to new daily attacks. Artificial intelligence-based methods have recently been used to secure and monitor network data. The present paper focuses on detecting and classifying multiple intrusions using logistic regression and deep convolutional neural networks (DCNN). The experimentations were carried out on benchmark UNSW-NB15 datasets, and the outcomes were validated with an enhanced success rate. The complex UNSW-NB15 data was initially renovated and then applied to the random forest model to select the most suitable features for training the logistic regression and DCNN models. The results show that the models used significantly enhanced the performance of detecting multiple network intrusions.