.. DO NOT EDIT. .. THIS FILE WAS AUTOMATICALLY GENERATED BY SPHINX-GALLERY. .. TO MAKE CHANGES, EDIT THE SOURCE PYTHON FILE: .. "generated/sphinx_gallery_examples/2-pool-regression/plot-GreedySamplingTarget-Improved_Greedy_Sampling_(GSi).py" .. LINE NUMBERS ARE GIVEN BELOW. .. only:: html .. note:: :class: sphx-glr-download-link-note :ref:`Go to the end ` to download the full example code. .. rst-class:: sphx-glr-example-title .. _sphx_glr_generated_sphinx_gallery_examples_2-pool-regression_plot-GreedySamplingTarget-Improved_Greedy_Sampling_(GSi).py: Improved Greedy Sampling (GSi) ============================== .. GENERATED FROM PYTHON SOURCE LINES 7-8 **Idea:** Improved Greedy Sampling (GSi) at first selects those samples that maximize the diversity in the feature space and then selects samples to maximize the diversity in the combined space of features and targets. The diversity is maximized by selecting those samples that are the furthest away from all previously labeled samples. The distance in the combined space is measured as the product of the distance in the feature space and the target space. To measure the distance in the target space the strategy predicts the label corresponding to each unlabeled sample. .. GENERATED FROM PYTHON SOURCE LINES 10-20 | **Google Colab Note**: If the notebook fails to run after installing the needed packages, try to restart the runtime (Ctrl + M) under Runtime -> Restart session. .. image:: https://colab.research.google.com/assets/colab-badge.svg :target: https://colab.research.google.com/github/scikit-activeml/scikit-activeml.github.io/blob/gh-pages/latest/generated/sphinx_gallery_notebooks//2-pool-regression/plot-GreedySamplingTarget-Improved_Greedy_Sampling_(GSi).ipynb | **Notebook Dependencies** | Uncomment the following cell to install all dependencies for this tutorial. .. GENERATED FROM PYTHON SOURCE LINES 20-23 .. code-block:: Python # !pip install scikit-activeml .. GENERATED FROM PYTHON SOURCE LINES 24-129 .. code-block:: Python import numpy as np from matplotlib import pyplot as plt, animation from scipy.stats import uniform from skactiveml.utils import MISSING_LABEL, is_labeled from skactiveml.regressor import NICKernelRegressor from skactiveml.pool import GreedySamplingTarget # Set a fixed random state for reproducibility. random_state = np.random.RandomState(0) def true_function(X_): """Compute the true underlying function.""" return (X_**3 + 2 * X_**2 + X_ - 1).flatten() # Generate samples. n_samples = 100 X = np.concatenate( [ uniform.rvs(0, 1.5, 9 * n_samples // 10, random_state=random_state), uniform.rvs(1.5, 0.5, n_samples // 10, random_state=random_state), ] ).reshape(-1, 1) # Define noise: higher noise for X < 1 and lower otherwise. noise = np.vectorize( lambda x: random_state.rand() * 1.5 if x < 1 else random_state.rand() * 0.5 ) # Build the dataset. y_true = true_function(X) + noise(X).flatten() y = np.full(shape=y_true.shape, fill_value=MISSING_LABEL) X_test = np.linspace(0, 2, num=100).reshape(-1, 1) y_test = true_function(X_test) # Initialise the regressor. reg = NICKernelRegressor(random_state=random_state, metric_dict={'gamma': 15.0}) # Initialise the query strategy. qs = GreedySamplingTarget(method='GSi') # Prepare the plotting area. fig, (ax_1, ax_2) = plt.subplots(2, 1, sharex=True) artists = [] # Active learning cycle. n_cycles = 20 for c in range(n_cycles): # Fit the regressor using the current labels. reg.fit(X, y) # Query the next sample(s). query_idx = qs.query(X=X, y=y, reg=reg) # Record current plot elements. coll_old = list(ax_1.collections) + list(ax_2.collections) title = ax_1.text( 0.5, 1.05, f"Prediction after acquiring {c} labels\n" f"Test R-squared score: {reg.score(X_test, y_test):.4f}", size=plt.rcParams["axes.titlesize"], ha="center", transform=ax_1.transAxes, ) ax_1.set_xlabel('Sample') ax_1.set_ylabel('Target Value') ax_2.set_xlabel('Sample') ax_2.set_ylabel('Utility') # Compute utility values for the test candidates. _, utilities_test = qs.query(X=X, y=y, reg=reg, candidates=X_test, return_utilities=True) utilities_test = (utilities_test - utilities_test.min()).flatten() if np.any(utilities_test != utilities_test[0]): utilities_test /= utilities_test.max() # Plot utility information on the second axis. (utility_line,) = ax_2.plot(X_test, utilities_test, c="green") utility_fill = plt.fill_between(X_test.flatten(), utilities_test, color="green", alpha=0.3) # Plot the samples and their labels. is_lbld = is_labeled(y) ax_1.scatter(X[~is_lbld], y_true[~is_lbld], c="lightblue") ax_1.scatter(X[is_lbld], y[is_lbld], c="orange") # Predict and plot the regressor's output. y_pred = reg.predict(X_test) (prediction_line,) = ax_1.plot(X_test, y_pred, c="black") # Capture new plot elements. coll_new = list(ax_1.collections) + list(ax_2.collections) coll_new.append(title) artists.append( [x for x in coll_new if (x not in coll_old)] + [utility_line, utility_fill, prediction_line] ) # Update labels for the queried sample. y[query_idx] = y_true[query_idx] # Create an animation from the collected frames. ani = animation.ArtistAnimation(fig, artists, interval=1000, blit=True) .. container:: sphx-glr-animation .. raw:: html
.. GENERATED FROM PYTHON SOURCE LINES 130-131 .. image:: ../../examples/pool_regression_legend.png .. GENERATED FROM PYTHON SOURCE LINES 133-138 .. rubric:: References: The implementation of this strategy is based on :footcite:t:`wu2019active`. .. footbibliography:: .. rst-class:: sphx-glr-timing **Total running time of the script:** (0 minutes 8.266 seconds) .. _sphx_glr_download_generated_sphinx_gallery_examples_2-pool-regression_plot-GreedySamplingTarget-Improved_Greedy_Sampling_(GSi).py: .. only:: html .. container:: sphx-glr-footer sphx-glr-footer-example .. container:: sphx-glr-download sphx-glr-download-jupyter :download:`Download Jupyter notebook: plot-GreedySamplingTarget-Improved_Greedy_Sampling_(GSi).ipynb ` .. container:: sphx-glr-download sphx-glr-download-python :download:`Download Python source code: plot-GreedySamplingTarget-Improved_Greedy_Sampling_(GSi).py ` .. container:: sphx-glr-download sphx-glr-download-zip :download:`Download zipped: plot-GreedySamplingTarget-Improved_Greedy_Sampling_(GSi).zip ` .. only:: html .. rst-class:: sphx-glr-signature `Gallery generated by Sphinx-Gallery `_