Hands-On Machine Learning for Algorithmic Trading

Forward propagation

The forward_prop function combines the previous operations to yield the output activations from the input data as a function of weights and biases, as follows:

def forward_prop(data, hidden_weights, hidden_bias, output_weights,            output_bias):    """Neural network as function."""    hidden_activations = hidden_layer(data, hidden_weights, hidden_bias)    return output_layer(hidden_activations, output_weights, output_bias)

The predict function produces the binary class predictions given weights, biases, and input data, as follows:

def predict(data, hidden_weights, hidden_bias, output_weights, output_bias):    """Predicts class 0 or 1"""    y_pred_proba = forward_prop(data,        q                        hidden_weights,                                hidden_bias,                                output_weights,                                output_bias) return ...