Neural Neural Network Blackbox Visualization and Modeling using Simulink
nn_sample.mat contains the input (ts), weights, and biases for the neural network to work.
Step 1: On the MATLAB command window type in,
load nn_sample.mat
Make sure that you are on the path where the .mat file is located.
Step 2: Double click on the nn_blackbox.slx project.
Step 3: Set the Simulation end time to 1s.
Step 4: Run the simulation.
Step 5: Observe the output on the scope.
This simple neural network used predetermined weights and biases. I will update the model to make it self-learning. For now, this just aims to visualize what happens inside the "blackbox" of a neural network.
About the model:
The model consists of dosage inputs from 0 to 1. The idea is to determine whether a particular value of dosage is effective or not (binary classification). The model has one hidden layer and two hidden neurons. A rectified linear unit (ReLU) activation function is used. The weights and biases manipulates the shape of the activation function to fit the data.
The inputs, weights, and biases are parameterized.