Hands-On Machine Learning for Algorithmic Trading

Implementing momentum updates using Python

To incorporate momentum into the parameter updates, define an update_momentum function that combines the results of the preceding compute_gradients function with the most recent momentum updates as follows for each parameter matrix:

def update_momentum(X, y_true, param_list, Ms, momentum_term, eta):    """Compute updates with momentum."""    gradients = compute_gradients(X, y_true, *param_list)    return [momentum_term * momentum - eta * grads            for momentum, grads in zip(Ms, gradients)]

The update_params function performs the actual updates:

def update_params(param_list, Ms):    """Update the parameters."""    return [P + M for P, M in zip(param_list, Ms)]