Neural networks are composed of many small processors that work simultaneously on the same task. They can learn from training data and use their knowledge to compare patterns in a dataset. Combining the strengths of parallel processing and distributed computing, the neural network can enhance the processing times for both the learning and execution stages to efficiently calculate the most probable output with a remarkable degree of accuracy. The aim of this research was to design, implement, and demonstrate the enhanced computational speed of a generalized large-scale neural network with broad-based applications using parallel computing. As a result, we evaluated distributed computing systems and compared the performance of different neural network training functions to determine which one worked best for good system performance. The results indicate that the proposed method outperforms other methods in terms of accuracy and convergence time. The study suggests that parallelism of the backpropagation neural network
model can lead to faster training convergence time and higher accuracy of the results. The system takes input data and runs on different systems in parallel.