Active Learning

Sometimes, we may just want to learn a black-box function, rather than optimizing it. This goal is known as active learning and corresponds to choosing query points that reduce our model uncertainty. This notebook demonstrates how to perform Bayesian active learning using Trieste.

%matplotlib inline
import numpy as np
import tensorflow as tf

np.random.seed(1793)
tf.random.set_seed(1793)

Describe the problem

In this example, we will perform active learning for the scaled Branin function.

from trieste.objectives import ScaledBranin
from trieste.experimental.plotting import plot_function_plotly

scaled_branin = ScaledBranin.objective
search_space = ScaledBranin.search_space

fig = plot_function_plotly(
    scaled_branin,
    search_space.lower,
    search_space.upper,
)
fig.show()