File size: 4,261 Bytes
5322e36 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 |
import numpy as np
import matplotlib.pyplot as plt
from PIL import Image
from io import BytesIO
from sklearn.linear_model import Ridge
from sklearn.gaussian_process.kernels import ExpSineSquared
from sklearn.kernel_ridge import KernelRidge
import gradio as gr
import random
def generate_data(n_samples: int) -> tuple:
rng = np.random.RandomState(random.randint(0, 1000))
data = np.linspace(0, 30, num=n_samples).reshape(-1, 1)
target = np.sin(data).ravel()
training_sample_indices = rng.choice(
np.arange(0, int(0.4 * n_samples)), size=int(0.2 * n_samples), replace=False
)
training_data = data[training_sample_indices]
training_noisy_target = (
target[training_sample_indices] + 0.5 * rng.randn(len(training_sample_indices))
)
return data, target, training_data, training_noisy_target
# def plot_ridge_and_kernel(n_samples: int) -> plt.figure:
# data, target, training_data, training_noisy_target = generate_data(n_samples)
# ridge = Ridge().fit(training_data, training_noisy_target)
# kernel_ridge = KernelRidge(kernel=ExpSineSquared())
# kernel_ridge.fit(training_data, training_noisy_target)
# fig, ax = plt.subplots(figsize=(8, 4))
# ax.plot(data, target, label="True signal", linewidth=2)
# ax.scatter(
# training_data,
# training_noisy_target,
# color="black",
# label="Noisy measurements",
# )
# ax.plot(data, ridge.predict(data), label="Ridge regression")
# ax.plot(
# data,
# kernel_ridge.predict(data),
# label="Kernel ridge",
# linewidth=2,
# linestyle="dashdot",
# )
# ax.legend()
# ax.set_xlabel("data")
# ax.set_ylabel("target")
# _ = ax.set_title("Ridge vs Kernel Ridge")
# return fig
def plot_ridge_and_kernel(n_samples: int) -> plt.figure:
data, target, training_data, training_noisy_target = generate_data(n_samples)
ridge = Ridge().fit(training_data, training_noisy_target)
kernel_ridge = KernelRidge(kernel=ExpSineSquared())
kernel_ridge.fit(training_data, training_noisy_target)
fig, ax = plt.subplots(figsize=(8, 4))
ridge_predictions = ridge.predict(data)
kernel_ridge_predictions = kernel_ridge.predict(data)
ax.plot(data, target, label="True signal", linewidth=2)
ax.scatter(
training_data,
training_noisy_target,
color="black",
label="Noisy measurements",
)
ax.plot(data, ridge_predictions, label="Ridge regression")
ax.plot(
data,
kernel_ridge_predictions,
label="Kernel ridge",
linewidth=2,
linestyle="dashdot",
)
ax.fill_between(
data.ravel(),
ridge_predictions,
kernel_ridge_predictions,
color="lightgrey",
alpha=0.4,
)
ax.legend()
ax.set_xlabel("data")
ax.set_ylabel("target")
_ = ax.set_title("Ridge vs Kernel Ridge with the area between highlighted")
return fig
def gradio_plot(n_samples: int) -> Image.Image:
fig = plot_ridge_and_kernel(n_samples)
buf = BytesIO()
fig.savefig(buf, format="png")
buf.seek(0)
im = Image.open(buf)
return im
inputs = [
gr.inputs.Slider(minimum=100, maximum=5000, step=100, label="n_samples", default=1000),
]
# Create the Gradio app
title = "Comparison of kernel ridge and Gaussian process regression"
description = "Kernel ridge regression and Gaussian process regression both use the kernel trick to fit data. While kernel ridge regression aims to find a single target function minimizing the loss (mean squared error), Gaussian process regression takes a probabilistic approach, defining a Gaussian posterior distribution over target functions using Bayes' theorem. Essentially, kernel ridge regression seeks one best function, while Gaussian process regression considers a range of probable functions based on prior probabilities and observed data. \n \n link to the official doc https://scikit-learn.org/stable/auto_examples/gaussian_process/plot_compare_gpr_krr.html#sphx-glr-auto-examples-gaussian-process-plot-compare-gpr-krr-py"
iface = gr.Interface(fn=gradio_plot, inputs=inputs, outputs="image", title = title , description = description)
iface.launch() |