Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,70 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import numpy as np
|
2 |
+
import matplotlib.pyplot as plt
|
3 |
+
import gradio as gr
|
4 |
+
from PIL import Image
|
5 |
+
|
6 |
+
from sklearn.gaussian_process import GaussianProcessClassifier
|
7 |
+
from sklearn.gaussian_process.kernels import RBF, DotProduct
|
8 |
+
|
9 |
+
|
10 |
+
def classify_xor_dataset(kernel_name):
|
11 |
+
xx, yy = np.meshgrid(np.linspace(-3, 3, 50), np.linspace(-3, 3, 50))
|
12 |
+
rng = np.random.RandomState(0)
|
13 |
+
X = rng.randn(200, 2)
|
14 |
+
Y = np.logical_xor(X[:, 0] > 0, X[:, 1] > 0)
|
15 |
+
|
16 |
+
# fit the model
|
17 |
+
fig, ax = plt.subplots(figsize=(10, 5))
|
18 |
+
kernels = [1.0 * RBF(length_scale=1.15), 1.0 * DotProduct(sigma_0=1.0) ** 2]
|
19 |
+
kernel_idx = 0 if kernel_name == "RBF" else 1
|
20 |
+
kernel = kernels[kernel_idx]
|
21 |
+
clf = GaussianProcessClassifier(kernel=kernel, warm_start=True).fit(X, Y)
|
22 |
+
|
23 |
+
# plot the decision function for each datapoint on the grid
|
24 |
+
Z = clf.predict_proba(np.vstack((xx.ravel(), yy.ravel())).T)[:, 1]
|
25 |
+
Z = Z.reshape(xx.shape)
|
26 |
+
|
27 |
+
ax.imshow(
|
28 |
+
Z,
|
29 |
+
interpolation="nearest",
|
30 |
+
extent=(xx.min(), xx.max(), yy.min(), yy.max()),
|
31 |
+
aspect="auto",
|
32 |
+
origin="lower",
|
33 |
+
cmap=plt.cm.PuOr_r,
|
34 |
+
)
|
35 |
+
ax.contour(xx, yy, Z, levels=[0.5], linewidths=2, colors=["k"])
|
36 |
+
ax.scatter(X[:, 0], X[:, 1], s=30, c=Y, cmap=plt.cm.Paired, edgecolors=(0, 0, 0))
|
37 |
+
ax.set_xticks(())
|
38 |
+
ax.set_yticks(())
|
39 |
+
ax.axis([-3, 3, -3, 3])
|
40 |
+
ax.set_title(
|
41 |
+
"%s\n Log-Marginal-Likelihood:%.3f"
|
42 |
+
% (clf.kernel_, clf.log_marginal_likelihood(clf.kernel_.theta)),
|
43 |
+
fontsize=12,
|
44 |
+
)
|
45 |
+
|
46 |
+
fig.canvas.draw()
|
47 |
+
pil_image = Image.frombytes("RGB", fig.canvas.get_width_height(), fig.canvas.tostring_rgb())
|
48 |
+
plt.close(fig)
|
49 |
+
return pil_image
|
50 |
+
|
51 |
+
|
52 |
+
title = "Gaussian Process Classification on the XOR Dataset"
|
53 |
+
description = "This example illustrates GPC on XOR data. Compared are a stationary, isotropic kernel (RBF) and a non-stationary kernel (DotProduct). On this particular dataset, the DotProduct kernel obtains considerably better results because the class-boundaries are linear and coincide with the coordinate axes. In general, stationary kernels often obtain better results. See the original scikit-learn example at https://scikit-learn.org/stable/auto_examples/gaussian_process/plot_gpc_xor.html"
|
54 |
+
kernel_options = ["RBF", "DotProduct"]
|
55 |
+
iface = gr.Interface(
|
56 |
+
classify_xor_dataset,
|
57 |
+
gr.inputs.Radio(choices=kernel_options, label="Kernel"),
|
58 |
+
gr.outputs.Image(label="Decision Boundary", type="pil"),
|
59 |
+
title=title,
|
60 |
+
description=description,
|
61 |
+
theme="default",
|
62 |
+
layout="vertical",
|
63 |
+
analytics_enabled=False,
|
64 |
+
examples=[
|
65 |
+
["RBF"],
|
66 |
+
["DotProduct"],
|
67 |
+
],
|
68 |
+
)
|
69 |
+
|
70 |
+
iface.launch()
|