Neural Networks - Model Representation

Neural Networks - Model Representation

Terminologies:

Neuron = logistic unit
$x_0$ = 1 = bias unit
$\Theta$ = parameter, weight
$h_{\theta} (x) = \frac{1}{1 + e^{-\theta^T x}}$ = sigmoid (logistic) activation function
Layer 1 (input layer) contains input units
Final layer that output $h_{\theta} (x)$ is output layer
The rest are hidden layers
$a_i^{(j)}$ = "activation" (value that's computed by and as output by previous layer) of unit i in layer j
$\Theta^{(j)}$ = matrix of weights controlling function mapping from layer j to layer j + 1

Resources:

- https://www.coursera.org/learn/machine-learning/supplement/Bln5m/model-representation-i