Robust least squares

Faisal Qureshi
Professor
Faculty of Science
Ontario Tech University
Oshawa ON Canada
http://vclab.science.ontariotechu.ca

Copyright information

© Faisal Qureshi

License

This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.

import numpy as np
import scipy as sp
import matplotlib.pyplot as plt
import matplotlib.lines as mlines
import matplotlib.cm as cm

Lesson Plan

• Robust least squares
• Outliers
• Loss functions
  • Linear loss
  • Soft L1 loss
  • Huber loss
  • Cauchy loss
  • arctan loss
  • Bisquare loss
• Incomplete data
• Mixed data

# Generate some data
np.random.seed(0)

m_true = .75
c_true = 1

x = np.hstack([np.linspace(3, 5, 5), np.linspace(7, 8, 5)])
y = m_true * x + c_true + np.random.rand(len(x))

# Outliers
y[9] = 1
y[0] = 9

data = np.vstack([x, y]).T
n = data.shape[0]

plt.figure(figsize=(7,7))
plt.scatter(data[:,0], data[:,1], c='red', alpha=0.9)
plt.xlim([0,10])
plt.ylim([0,10])
plt.title('Generating some points from a line');

x = data[:,0]
y = data[:,1]

print(f'x =\n{x.reshape(n,1)}')
print(f'y =\n{y.reshape(n,1)}')

x =
[[3.  ]
 [3.5 ]
 [4.  ]
 [4.5 ]
 [5.  ]
 [7.  ]
 [7.25]
 [7.5 ]
 [7.75]
 [8.  ]]
y =
[[9.        ]
 [4.34018937]
 [4.60276338]
 [4.91988318]
 [5.1736548 ]
 [6.89589411]
 [6.87508721]
 [7.516773  ]
 [7.77616276]
 [1.        ]]

theta = np.polyfit(x, y, 1)
m = theta[0]
c = theta[1]

print(f'c = {c}')
print(f'm = {m}')

c = 6.27194557802188
m = -0.08033126904046138

x_plotting = np.linspace(0,10,100)
y_plotting = m * x_plotting + c

plt.figure(figsize=(7,7))
plt.scatter(data[:,0], data[:,1], c='red', alpha=0.9)
plt.plot(x_plotting, y_plotting, 'b-')
plt.xlim([0,10])
plt.ylim([0,10])
plt.title('The effect of outliers on least squares');

Outliers

Outlier is a data point that differs significantly from the other observations. These can have large negative impact on model fitting. These are difficult to deal with, since we have no way of knowning a priori that a particular data item is an outlier.

# Generate some data
np.random.seed(0)

m_true = .75
c_true = 1

x = np.hstack([np.linspace(1, 5, 50), np.linspace(7, 10, 50)])
y = m_true * x + c_true + np.random.rand(len(x))*.2

data = np.vstack([x, y]).T
n = data.shape[0]

outlier_percentage = 35
ind = (np.random.rand(int(n*outlier_percentage/100))*n).astype('int')
data[ind, 1] = np.random.rand(len(ind))*100
#data[ind, 1] = np.random.rand(len(ind))*10+90

plt.figure(figsize=(7,7))
plt.scatter(data[:,0], data[:,1], c='red', alpha=0.9)
plt.xlim([-30,30])
plt.ylim([-30,30])
#plt.ylim([0,100])
plt.title('Noisy data');

def draw_line(theta, ax, **kwargs):
    m = theta[0]
    c = theta[1]
    
    xmin, xmax = ax.get_xbound()
    
    ymin = m * xmin + c
    ymax = m * xmax + c
        
    l = mlines.Line2D([xmin, xmax], [ymin,ymax], **kwargs)
    ax.add_line(l)

theta = np.polyfit(data[:,0], data[:,1], 1)
print(f'#data = {n}')
print(f'theta = {theta}')

#data = 100
theta = [1.8094878  8.29787401]

plt.figure(figsize=(7,7))
plt.scatter(data[:,0], data[:,1], c='red', alpha=0.9)
plt.xlim([-30,30])
plt.ylim([-30,30])
#plt.ylim([0,100])
plt.title('Least squares in the presence of outliers');
draw_line(theta, plt.gca(), c='black', linewidth='3')

Loss functions

In the discussion below, set $$ e = y_{\mathrm{prediction}} - y_{\mathrm{true}} $$ and $$ z = e^2 $$ In other words, $z$ is error-squared.

The following expressions are pulled out from scipy implementation of least squares. For an in depth discussion, I suggest "Convex Optimization" by Boyd et al, 2004 to the interested reader.

Linear loss

$$ \rho(z) = z $$

Soft L1 loss

$$ \rho(z) = 2 \left(\sqrt{1+z} - 1\right) $$

Huber loss

Generally

$$ \rho(e) = \begin{cases} \frac{1}{2} e^2 & \text{for } |e| \le k \\ k|e| - \frac{1}{2}k^2 & \text{otherwise} \end{cases} $$

Here $k$ is referred to as the tuning constant. Smaller values of $k$ produce more resistance to outliers, but at the expense of lower efficiency when the errors are normally distributed. The tuning constant is generally picked to give reasonably high efficiency in the normal case. Typically, we set $k = 1.345 \sigma$ for the Huber (where $\sigma$ is the standard deviation of the errors) produce 95-percent efficiency when the errors are normal, and still offer protection against outliers.

In scipy.optimize.least_square Huber loss is defined as

$$ \rho(z) = \begin{cases} z & \text{for } z \le 0 \\ 2 \left( \sqrt(1+z) - 1 \right) & \text{otherwise} \end{cases} $$

Cauchy loss

$$ \rho(z) = \ln (1 + z) $$

arctan loss

$$ \rho(z) = \arctan(z) $$

Bisquare

Generally

$$ \rho(e) = \begin{cases} \frac{k^2}{6} \{ 1 - \left[ 1 - \left( \frac{e}{k} \right)^2 \right]^3 \} & \text{for } |e| \le k \\ \frac{k^2}{6} & \text{otherwise} \end{cases} $$

$$ \rho(z) = \frac{z}{z + \sigma^2} $$

Here $\sigma$ is often referred to a scale parameter.

e = np.linspace(-3,3,100)
z = np.square(e)

linear_loss = z
l1_loss = np.abs(e)
soft_l1_loss = 2*(np.power(1+z, 0.5) - 1) # Smooth approx. of l1 loss.
huber_loss = 2*np.power(z, 0.5) - 1
huber_loss[z <= 1] = z[z <= 1]
cauchy_loss = np.log(1 + z)
arctan_loss = np.arctan(z)

scale = 0.1
robust_loss = z / (z + scale**2)

plt.figure(figsize=(15,15))
plt.subplot(331)
plt.title('Linear loss')
plt.ylim(0,10)
plt.grid()
plt.plot(e, linear_loss, label='Linear loss')
plt.subplot(332)
plt.grid()
plt.ylim(0,10)
plt.title('L1 loss')
plt.plot(e, l1_loss, label='L1 loss')
plt.subplot(333)
plt.grid()
plt.ylim(0,10)
plt.title('Soft L1 loss')
plt.plot(e, soft_l1_loss, label='Soft L1 loss')
plt.subplot(334)
plt.grid()
plt.ylim(0,10)
plt.title('Huber loss')
plt.plot(e, huber_loss, label='Huber loss')
plt.subplot(335)
plt.grid()
plt.ylim(0,10)
plt.title('Cauchy loss')
plt.plot(e, cauchy_loss, label='Cauchy loss')
plt.subplot(336)
plt.grid()
plt.ylim(0,10)
plt.title('Arctan loss')
plt.plot(e, arctan_loss, label='Arctan loss')
plt.subplot(337)
plt.grid()
plt.ylim(0,10)
plt.title('?Loss')
plt.plot(e, robust_loss, label='Robust loss')
plt.subplot(338)
plt.grid()
plt.title('?Loss (Rescaled y-axis)')
plt.plot(e, z / (z + 2**2), label='$\sigma=2$')
plt.plot(e, z / (z + 1**2), label='$\sigma=1$')
plt.plot(e, z / (z + .5**2), label='$\sigma=.5$')
plt.plot(e, z / (z + .25**2), label='$\sigma=.25$')
plt.plot(e, z / (z + .1**2), label='$\sigma=.1$')
plt.legend();

# Generate some data
np.random.seed(0)

m_true = .75
c_true = 1

x = np.hstack([np.linspace(1, 5, 50), np.linspace(7, 10, 50)])
y = m_true * x + c_true + np.random.rand(len(x))*.2

data = np.vstack([x, y]).T
n = data.shape[0]

outlier_percentage = 50
ind = (np.random.rand(int(n*outlier_percentage/100))*n).astype('int')
data[ind, 1] = np.random.rand(len(ind))*9

plt.figure(figsize=(7,7))
plt.scatter(data[:,0], data[:,1], c='red', alpha=0.9)
plt.xlim([0,10])
plt.ylim([0,10])
plt.title('Noisy data');

def line_fitting_errors(x, data):
    m = x[0]
    c = x[1]
    y_predicted = m * data[:,0] + c
    e = y_predicted - data[:,1]
    return e

theta_guess = [1, 1] # Need an intial guess

# Available losses
losses = [
    'linear',
    'soft_l1',
    'huber',
    'cauchy',
    'arctan'
]
loss_id = 4  # Pick a loss

# Non-linear least square fitting
from scipy.optimize import least_squares
retval = least_squares(line_fitting_errors, theta_guess, loss=losses[loss_id], args=[data])
print(f'Reasons for stopping: {retval["message"]}')
print(f'Success: {retval["success"]}')
theta = retval['x']
print(f'#data = {n}')
print(f'theta = {theta}')

Reasons for stopping: `ftol` termination condition is satisfied.
Success: True
#data = 100
theta = [0.73475407 1.15731305]

plt.figure(figsize=(7,7))
ax = plt.gca()
plt.scatter(data[:,0], data[:,1], c='red', alpha=0.9)
plt.xlim([0,10])
plt.ylim([0,10])
plt.title(f'Non linear least squares.  Loss = {losses[loss_id]}')
draw_line(theta, ax, c='black', linewidth='3');

Robust least squares

• Robust least square fitting is a non-linear optimization problem, which must be solved iterative.
• The iterative problem can be initialized using least square fitting.

Incomplete data or data from multiple sources

• How do we deal with incomplete data?
• What happen if we data is coming from two different (unrelated) models?

Incomplete data

# Generate some data
np.random.seed(0)

m_true = .75
c_true = 0
x = np.linspace(1, 8, 150)
y = m_true * x + c_true + np.random.rand(len(x))*3

data = np.vstack([x, y]).T
n = data.shape[0]

plt.figure(figsize=(7,7))
plt.scatter(x, y, c='red', alpha=0.9)
plt.xlim([0,10])
plt.ylim([0,10])
plt.title('Complete data');

cutoff = 5 
x_available = x[:cutoff]
y_available = y[:cutoff]

data = np.vstack([x_available, y_available]).T
n = data.shape[0]

plt.figure(figsize=(7,7))
plt.scatter(data[:,0], data[:,1], c='red', alpha=0.9)
plt.scatter(x[cutoff:], y[cutoff:], c='red', alpha=0.1)
plt.xlim([0,10])
plt.ylim([0,10])
plt.title('Available data');

Fitting model to available data

theta = np.polyfit(data[:,0], data[:,1], 1)
print(f'#available data = {n}')
print(f'theta = {theta}')

#available data = 5
theta = [-1.93598208  4.63953878]

plt.figure(figsize=(7,7))
plt.scatter(data[:,0], data[:,1], c='red', alpha=0.9)
plt.scatter(x[cutoff:], y[cutoff:], c='red', alpha=0.1)
plt.xlim([0,10])
plt.ylim([0,10])
plt.title('Modeling fitting to available data');
draw_line(theta, plt.gca(), c='black', linewidth='3')

Mixed data

# Generate some data
np.random.seed(0)

m_true_1 = .75
c_true_1 = 1
x1 = np.linspace(1, 8, 150)
y1 = m_true_1 * x1 + c_true_1 + np.random.rand(len(x1))*2

m_true_2 = -.5
c_true_2 = 8
x2 = np.linspace(4, 9.5, 20)
y2 = m_true_2 * x2 + c_true_2 + np.random.rand(len(x2))*1.4

x = np.hstack([x1, x2])
y = np.hstack([y1, y2])

data = np.vstack([x, y]).T
n = data.shape[0]

plt.figure(figsize=(7,7))
plt.scatter(x1, y1, c='red', alpha=0.9)
plt.scatter(x2, y2, c='blue', alpha=0.5)
plt.xlim([0,10])
plt.ylim([0,10])
plt.title('Noisy data');

theta = np.polyfit(data[:,0], data[:,1], 1)
print(f'#data = {n}')
print(f'theta = {theta}')

#data = 170
theta = [0.58011192 2.62702126]

plt.figure(figsize=(7,7))
ax = plt.gca()
plt.scatter(data[:,0], data[:,1], c='red', alpha=0.9)
plt.xlim([0,10])
plt.ylim([0,10])
plt.title('Least squares in the presence of outliers');
draw_line(theta, ax, c='black', linewidth='3')

Summary

• Robust least square fitting
  • A non-linear optimization problem
• The importance of choosing appropriate losses
• Data related issues
  • Outliers
  • Incomplete data
  • Mixed data