Experimenting with hyperparameters with Deeplearning4j

Let's build a simple neural network to demonstrate the effects of various hyperparameters on model performance. We will create a simple neural network that can add two numbers based on the randomly generated training data. The training data has two independent variables, x1 and x2, and an output variable, y1 = x1 + x2. Here is a pictorial view of the network we will generate with the deeplearning4j library:

Here is the utility code for generating the sample data: x1 and x2 as the input independent variables, and y as the output (dependent) variable:

// Method to generate the training data ...

Get Artificial Intelligence for Big Data now with O’Reilly online learning.

O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.