"five random points to initialise things, then five using the GP model"

[Gilles Louppe]

Just read through the code.

To achieve "five random points to initialise things, then five using the GP model" in the bayesian optimization loop, one needs to set n_random_starts=5 and n_calls=10, and not n_calls=5 as currently performed in the notebook.

Also, I would recommend using many more calls, as 10 iterations is quite small.

[Gilles Louppe]

Fixed from our discussions and last changes.