Neural Networks and Deep Learning
Richard E. Neapolitan, Xia Jiang · 2018
Neural networks have been used effectively in applications such as image recognition and speech recognition, which are hard to model with the structured approach used in rule-based systems and Bayesian networks. Bayesian networks, on the other hand, provide a relationship among variables, which can often be interpreted as causal. Furthermore, Bayesian networks enable to model and understand complex human decisions. A neural network consists of one to many artificial neurons, which communicate with each other. Bayesian networks, on the other hand, have more often been applied successfully to problems that involve determining the relationships among related random variables, and exploiting these relationships to do inference and make decisions. The output of one neuron is the input to another neuron. The chapter shows that SAT scores and parental income have an affect on whether a student graduates college. It constructs the neural network, and assigns values to the weights to achieve desired results.