Distinguishing two groups with a Neural Network

less than 1 minute read

Missing animation

Generate the dataset as described.
Label the data to indicate which points belong to the inner and outer groups.
Create and train the neural network.
Animate the learning process to show how the network’s decision boundary evolves.

The data is generated by two uniform distribution. The angles are drawn from $U(0, 2\pi)$ while the radii for the inner circle are drawn from $U(0, r_1)$ the outer radii are drawn from $U(r_2, r_3)$, with $r_1 < r_2 <r_3$. $x = r\cos(\alpha)$ $y = r\sin(\alpha)$

The neural network sequence is given by a tensorflow keras model with sixteen neurons in two layers using the rectifier activation function and a sigmoid for the output neuron. This sequence is chosen to account for the non-linearity of the two distinct groups.

model = Sequential(
    [
        Dense(16, input_dim=2, activation="relu"),
        Dense(16, activation="relu"),
        Dense(1, activation="sigmoid"),
    ]
)

Philipp Herz