Experimentation-on-Neural-Networks-with-MNIST-Handwritten-Digits-Images

The focus of this study is to understand the composition of a simple single-hidden layer neural network, primarily on how different numbers of nodes would affect the performance of the neural network. Understanding this effect would allow further exploration of deeper neural networks that have multi-hidden layers. This exercise utilized a Jupyter Lab instance and launched the Keras library for TensorFlow 2.0 with Python. Keras is a popular deep learning API for the development of machine learning models such as computer vision and natural language processing. Keras also allows the utilization of GPUs when training neural network models. During the experiments, the Python codes are executed to develop the models.