×
Manual
The leftmost neurons are input neurons and rightmost are output ones. The dataset is generated using
functions you set.
Network Configuration: Specify the number of neurons in each layer, e.g., "3,4,2" for 3
input, 4 hidden, and 2 output neurons.
Activation Function: Choose the activation function (Sigmoid, ReLU, Tanh, or Leaky ReLU)
used for the neurons.
Loss Function: Choose the loss function for training (Mean Squared Error, Cross Entropy, or
Binary Cross Entropy).
Input Values: Enter the input values in a function format (e.g., `{i} + {t}`). t is time and
i is index of neuron on input layer. You can use javascript within curlies.
Target Values: Enter the target values in a function format (e.g., `{i} + {t}`). t is time
and i is index of neuron on input layer. You can use javascript within curlies.
Buttons:
- Plot: Visualizes the neural network with the specified configuration.
- Start: Starts the training process.
- Stop: Stops the ongoing training process.
- Reset: Resets the neural network.
Explanation
Nodes (Neurons): The color of each node represents its activation value.
Green indicates a strong positive activation, red indicates a negative activation, and neutral or no color
suggests low or zero activation.
Weights (Connections): The color of the weights represents their magnitude and
sign. Green lines indicate positive weights (amplifying the signal), while red lines represent
negative weights (dampening the signal). The intensity of the color reflects the strength of the connection.