The major source of any social media to improve their promotions or sales is to construct a strong network that has more influence nodes. There are cases where the network is big and but the influence of the nodes has very less influence. Our previous work has implemented the enhanced frog leaping algorithm which has the drawback in terms of execution time and affinity values with the increase of network. The proposed work in this paper utilizes an ensemble neural network that can handle high-dimensionality data at a faster rate. Since the influence maximization is a connected graph, to identify the similarities among the connected nodes, the statistical approaches fail because it has to compare n-1 pairs. The neural networks use the concept of weights and bias the model trains and extract the similarity features and classify the necessary nodes using the last layer of the neural network. Designing all the layers of the network as dense will help the model to reduce the execution time with more influence nodes. The best estimators for each layer are identified by applying the Cuckoo Search, genetic approach. Integration of genetic algorithms with neural networks has reduced the time to identify the influence nodes. It also reduces the memory utilization to store the connectivity nodes because the general tendency of the neural network is to connect the nodes with dot products between them. When compared to the genetic approach, the execution time of the proposed model is reduced by nearly 6%.