How Many Hidden Units Should A Neural Network Have?

We will be discussing this interesting topic and others just like it at Boston's top Data Science Conference, ODSC East. Neural networks (NN) are powerful statistical models that can integrate lots of different kinds of data in complex ways. They are at their most powerful when their structure is optimally suited to the data you have available. NNs are composed of an input layer, hidden layers, and an output layer. 'Network architecture' refers to how many hidden layers a NN has, the number of units on each layer, and the patterns though which units are connected.