Description
1. [18 points] Explain the following concepts:
1) supervised learning,
2) unsupervised learning,
3) online learning,
4) batch learning,
5) model-based learning,
6) instance-based learning.
Programming Problem:
2. [42 points] In this problem, we write a program to estimate the parameters for an unknown
polynomial using the polyfit() function of the numpy package.
1) Please plot the noisy data and the polynomial you found (in the same figure). You can use any
value of m selected from 2, 3, 4, 5, 6.
2) Plot MSE versus order m for m = 1, 2, 3, 4, 5, 6, 7, 8. Identify the best choice of order m.
3) Change variable noise_scale to 150, 200, 400, 600, 1000 respectively, re-run the algorithm and
plot the polynomials with the m found in 2). Discuss the impact of noise scale to the accuracy of
the returned parameters. [You need to plot a figure like in 1) for each choice of noise_scale.]
4) Change variable number_of_samples to 40, 30, 20, 10 respectively, re-ran the algorithm and
plot the polynomials with the m found in 2). Discuss the impact of the number of samples to the
accuracy of the returned parameters. [You need to plot a figure like in 1) for each choice of
number_of_samples.]
A simulated dataset will be provided as below. The polynomial used is y = 5 * x + 20 * x2 + x3
.
Simulated data is given as follows in Python:
import matplotlib.pyplot as plt
plt.style.use(‘seaborn-whitegrid’)
import numpy as np
noise_scale = 100
number_of_samples = 50
x = 25*(np.random.rand(number_of_samples, 1) – 0.8)
y = 5 * x + 20 * x**2 + 1 * x**3 + noise_scale*np.random.randn(number_of_samples, 1)
plt.plot(x,y,’ro’)
