Intel® Extension for Scikit-learn TSNE example

from timeit import default_timer as timer
from sklearn import metrics
from sklearn.datasets import make_blobs
import matplotlib.pyplot as plt

%matplotlib inline

import warnings

warnings.filterwarnings("ignore")

Generate the data

Generate isotropic Gaussian blobs for clustering. With the number of samples: 20k Number of features: 100 Number of blobs: 4 Source:

https://scikit-learn.org/stable/modules/generated/sklearn.datasets.make_blobs.html

x, y = make_blobs(n_samples=20000, centers=4, n_features=100, random_state=0)

Patch original Scikit-learn with Intel® Extension for Scikit-learn

Intel® Extension for Scikit-learn (previously known as daal4py) contains drop-in replacement functionality for the stock Scikit-learn package. You can take advantage of the performance optimizations of Intel® Extension for Scikit-learn by adding just two lines of code before the usual Scikit-learn imports:

from sklearnex import patch_sklearn

patch_sklearn()

Intel(R) Extension for Scikit-learn* enabled (https://github.com/intel/scikit-learn-intelex)

Intel® Extension for Scikit-learn patching affects performance of specific Scikit-learn functionality. Refer to the list of supported algorithms and parameters for details. In cases when unsupported parameters are used, the package fallbacks into original Scikit-learn. If the patching does not cover your scenarios, submit an issue on GitHub.

Training TSNE algorithm with Intel® Extension for Scikit-learn for generated dataset

from sklearn.manifold import TSNE

params = {"n_components": 2, "random_state": 42}
start = timer()
tsne = TSNE(**params)
embedding_intelex = tsne.fit_transform(x)
time_opt = timer() - start

print(f"Intel® extension for Scikit-learn time: {time_opt:.2f} s")
print(f"Intel® Extension for scikit-learn. Divergence: {tsne.kl_divergence_}")

Intel® extension for Scikit-learn time: 12.63 s
Intel® Extension for scikit-learn. Divergence: 4.289110606110757

### Train the same algorithm with original Scikit-learn In order to cancel optimizations, we use unpatch_sklearn and reimport the class TSNE.

from sklearnex import unpatch_sklearn

unpatch_sklearn()

Training algorithm with original Scikit-learn library for generated dataset

from sklearn.manifold import TSNE

params = {"n_components": 2, "random_state": 42}
start = timer()
tsne = TSNE(**params)
embedding_original = tsne.fit_transform(x)
time_original = timer() - start

print(f"Original Scikit-learn time: {time_original:.2f} s")
print(f"Original Scikit-learn. Divergence: {tsne.kl_divergence_}")

Original Scikit-learn time: 37.66 s
Original Scikit-learn. Divergence: 4.2955403327941895

Plot embeddings original scikit-learn and Intel® extension

colors = [int(m) for m in y]

for emb, title in zip(
    [embedding_intelex, embedding_original],
    ["Intel® Extension for scikit-learn", "Original scikit-learn"],
):
    plt.scatter(emb[:, 0], emb[:, 1], c=colors)
    plt.title(title)
    plt.xlabel("x")
    plt.ylabel("y")
    plt.show()

f"Speedup for this run: {(time_original/time_opt):.1f}"

'Speedup for this run: 3.0'