Dimensionality Reduction

• PCA, ICA, LLE, Isomap,

Autoencoder
• PCA is the most important technique to
know. It takes advantage of correlations in
data dimensions to produce the best possible
lower dimensional representation based on
linear projections (minimizes reconstruction
error).

• PCA should be used for dimensionality

reduction, not for discovering patterns or
making predictions. Don't try to assign
semantic meaning to the bases.
• http://fakeisthenewreal.org/reform/
• http://fakeisthenewreal.org/reform/
Clustering example: image segmentation
Goal: Break up the image into meaningful or perceptually
similar regions
Segmentation for feature support or efficiency

50x50
Patch
50x50
Patch

[Felzenszwalb and Huttenlocher 2004]

[Shi and Malik 2001]

[Hoiem et al. 2005, Mori 2005] Slide: Derek Hoiem
Segmentation as a result

Rother et al. 2004

Types of segmentations

Oversegmentation Undersegmentation

Multiple Segmentations
Clustering: group together similar points and represent them
with a single token

Key Challenges:
1) What makes two points/images/patches similar?
2) How do we compute an overall grouping from pairwise similarities?

Slide: Derek Hoiem

How do we cluster?

• K-means
– Iteratively re-assign points to the nearest cluster center
• Agglomerative clustering
– Start with each point as its own cluster and iteratively merge the closest
clusters
• Mean-shift clustering
– Estimate modes of pdf
• Spectral clustering
– Split the nodes in a graph based on assigned links with similarity weights
Clustering for Summarization
Goal: cluster to minimize variance in data given clusters
– Preserve information

Cluster center Data

c * , δ *  argmin N1  ij c i  x j 
N K
2

c ,δ j i

Whether xj is assigned to ci

Slide: Derek Hoiem

K-means algorithm

1. Randomly
select K centers

2. Assign each
point to nearest
center

3. Compute new
center (mean)
for each cluster

Illustration: http://en.wikipedia.org/wiki/K-means_clustering
K-means algorithm

1. Randomly
select K centers

2. Assign each
point to nearest
center

Back to 2
3. Compute new
center (mean)
for each cluster

Illustration: http://en.wikipedia.org/wiki/K-means_clustering
K-means
1. Initialize cluster centers: c0 ; t=0

2. Assign each point to the closest center

 c x j
N K
t 1 2
δ  argmin
t 1
N ij i
δ j i

3. Update cluster centers as the mean of the points

 c 
N K
c  argmin x j
t 1 t 2
N ij i
c j i

4. Repeat 2-3 until no points are re-assigned (t=t+1)

Slide: Derek Hoiem
K-means converges to a local minimum
K-means: design choices

• Initialization
– Randomly select K points as initial cluster center
– Or greedily choose K points to minimize residual

• Distance measures
– Traditionally Euclidean, could be others

• Optimization
– Will converge to a local minimum
– May want to perform multiple restarts
K-means clustering using intensity or color

Image Clusters on intensity Clusters on color

How to evaluate clusters?

• Generative
– How well are points reconstructed from the clusters?

• Discriminative
– How well do the clusters correspond to labels?
• Purity
– Note: unsupervised clustering does not aim to be discriminative

Slide: Derek Hoiem

How to choose the number of clusters?
• Validation set
– Try different numbers of clusters and look at performance
• When building dictionaries (discussed later), more clusters typically work
better

Slide: Derek Hoiem

 K-Means pros and cons
• Pros
• Finds cluster centers that minimize
conditional variance (good
representation of data)
• Simple and fast*
• Easy to implement
• Cons
• Need to choose K
• Sensitive to outliers
• Prone to local minima
• All clusters have the same parameters
(e.g., distance measure is non-
adaptive)
• *Can be slow: each iteration is O(KNd)
for N d-dimensional points
• Usage
• Rarely used for pixel segmentation
Building Visual Dictionaries
1. Sample patches from
a database
– E.g., 128 dimensional
SIFT vectors

2. Cluster the patches

– Cluster centers are
the dictionary

3. Assign a codeword
(number) to each
new patch, according
to the nearest cluster
Examples of learned codewords

Most likely codewords for 4 learned “topics”

http://www.robots.ox.ac.uk/~vgg/publications/papers/sivic05b.pdf Sivic et al. ICCV 2005

Agglomerative clustering
Agglomerative clustering
Agglomerative clustering
Agglomerative clustering
Agglomerative clustering
Agglomerative clustering
How to define cluster similarity?
- Average distance between points, maximum
distance, minimum distance
- Distance between means or medoids

How many clusters?

- Clustering creates a dendrogram (a tree)
- Threshold based on max number of clusters
or based on distance between merges

distance
Conclusions: Agglomerative Clustering
Good
• Simple to implement, widespread application
• Clusters have adaptive shapes
• Provides a hierarchy of clusters

Bad
• May have imbalanced clusters
• Still have to choose number of clusters or threshold
• Need to use an “ultrametric” to get a meaningful hierarchy
Mean shift segmentation
D. Comaniciu and P. Meer, Mean Shift: A Robust Approach toward Feature Space Analysis, PAMI 2002.

• Versatile technique for clustering-based

segmentation
Mean shift algorithm
• Try to find modes of this non-parametric density
Kernel density estimation

Kernel density estimation function

Gaussian kernel
Mean shift
Region of
interest

Center of
mass

Mean Shift
vector

Slide by Y. Ukrainitz & B. Sarel

Mean shift
Region of
interest

Center of
mass

Mean Shift
vector

Slide by Y. Ukrainitz & B. Sarel

Mean shift
Region of
interest

Center of
mass

Mean Shift
vector

Slide by Y. Ukrainitz & B. Sarel

Mean shift
Region of
interest

Center of
mass

Mean Shift
vector

Slide by Y. Ukrainitz & B. Sarel

Mean shift
Region of
interest

Center of
mass

Mean Shift
vector

Slide by Y. Ukrainitz & B. Sarel

Mean shift
Region of
interest

Center of
mass

Mean Shift
vector

Slide by Y. Ukrainitz & B. Sarel

Mean shift
Region of
interest

Center of
mass

Slide by Y. Ukrainitz & B. Sarel

Computing the Mean Shift

Simple Mean Shift procedure:

• Compute mean shift vector

•Translate the Kernel window by m(x)

 n  x - xi 2  
  xi g   
 i 1  h  
m ( x)      x
 n  x - xi 
2

  g  h  
 i 1   
g(x)   k ( x)
Slide by Y. Ukrainitz & B. Sarel
 Attraction basin

• Attraction basin: the region for which all

trajectories lead to the same mode
• Cluster: all data points in the attraction
basin of a mode

Slide by Y. Ukrainitz & B. Sarel

Attraction basin
Mean shift clustering
• The mean shift algorithm seeks modes of the given set of
points
1. Choose kernel and bandwidth
2. For each point:
a) Center a window on that point
b) Compute the mean of the data in the search window
c) Center the search window at the new mean location
d) Repeat (b,c) until convergence
3. Assign points that lead to nearby modes to the same cluster
 Segmentation by Mean Shift
• Compute features for each pixel (color, gradients, texture, etc)
• Set kernel size for features Kf and position Ks
• Initialize windows at individual pixel locations
• Perform mean shift for each window until convergence
• Merge windows that are within width of Kf and Ks
Mean shift segmentation results

Comaniciu and Meer 2002

Comaniciu and Meer 2002
Mean shift pros and cons
• Pros
– Good general-practice segmentation
– Flexible in number and shape of regions
– Robust to outliers
• Cons
– Have to choose kernel size in advance
– Not suitable for high-dimensional features
• When to use it
– Oversegmentatoin
– Multiple segmentations
– Tracking, clustering, filtering applications
Which algorithm to try first?
• Quantization/Summarization: K-means
– Aims to preserve variance of original data
– Can easily assign new point to a cluster

Summary of 20,000 photos of Rome using

Quantization for
“greedy k-means”
computing histograms
http://grail.cs.washington.edu/projects/canonview/
The machine learning framework

• Apply a prediction function to a feature representation of

the image to get the desired output:

f( ) = “apple”
f( ) = “tomato”
f( ) = “cow”
Slide credit: L. Lazebnik
The machine learning framework

y = f(x)
output prediction Image
function feature

• Training: given a training set of labeled examples {(x1,y1),

…, (xN,yN)}, estimate the prediction function f by minimizing
the prediction error on the training set
• Testing: apply f to a never before seen test example x and
output the predicted value y = f(x)
Slide credit: L. Lazebnik
Learning a classifier
Given some set of features with corresponding labels, learn a
function to predict the labels from the features

x
x
x x x
x
o x
x
o o
o
o
x2

x1
 Steps
Training Training
Labels
Training
Images
Image Learned
Training
Features model

Testing

Image Learned
Prediction
Features model
Test Image Slide credit: D. Hoiem and L. Lazebnik
 Features
• Raw pixels

• Histograms

• GIST descriptors

• … Slide credit: L. Lazebnik

One way to think about it…

• Training labels dictate that two examples are the same or

different, in some sense

• Features and distance measures define visual similarity

• Classifiers try to learn weights or parameters for features and

distance measures so that visual similarity predicts label
similarity
Many classifiers to choose from
• SVM
• Neural networks
Which is the best one?
• Naïve Bayes
• Bayesian network
• Logistic regression
• Randomized Forests
• Boosted Decision Trees
• K-nearest neighbor
• RBMs
• Deep Convolutional Network
• Etc.
 Claim:
The decision to use machine learning
is more important than the choice of
a particular learning method.

*Deep learning seems to be an exception to this, at

the moment, probably because it is learning the
feature representation.
 Classifiers: Nearest neighbor

Test Training
Training examples
examples example
from class 2
from class 1

f(x) = label of the training example nearest to x

• All we need is a distance function for our inputs

• No training required!
Slide credit: L. Lazebnik
 Classifiers: Linear

• Find a linear function to separate the classes:

f(x) = sgn(w  x + b)

Slide credit: L. Lazebnik

Recognition task and supervision
• Images in the training set must be annotated with the
“correct answer” that the model is expected to produce

Contains a motorbike

Slide credit: L. Lazebnik

Unsupervised “Weakly” supervised Fully supervised

Definition depends on task

Slide credit: L. Lazebnik
 Generalization

Training set (labels known) Test set (labels

unknown)

• How well does a learned model generalize from

the data it was trained on to a new test set?
Slide credit: L. Lazebnik
 Generalization
• Components of generalization error
– Bias: how much the average model over all training sets differ
from the true model?
• Error due to inaccurate assumptions/simplifications made by
the model. “Bias” sounds negative. “Regularization” sounds
nicer.
– Variance: how much models estimated from different training
sets differ from each other.
• Underfitting: model is too “simple” to represent all the
relevant class characteristics
– High bias (few degrees of freedom) and low variance
– High training error and high test error
• Overfitting: model is too “complex” and fits irrelevant
characteristics (noise) in the data
– Low bias (many degrees of freedom) and high variance
– Low training error and high test error
Slide credit: L. Lazebnik
Bias-Variance Trade-off

• Models with too few

parameters are
inaccurate because of a
large bias (not enough
flexibility).

• Models with too many

parameters are
inaccurate because of a
large variance (too much
sensitivity to the sample).

Slide credit: D. Hoiem

 Bias-variance tradeoff

Underfitting Overfitting

Error

Test error

Training error

High Bias Low Bias

Low Variance
Complexity High Variance

Slide credit: D. Hoiem

 Bias-variance tradeoff

Test Error Few training examples

Many training examples

High Bias Low Bias

Low Variance
Complexity High Variance

Slide credit: D. Hoiem

 Effect of Training Size

Fixed prediction model

Error

Testing
Generalization Error

Training
Number of Training Examples

Slide credit: D. Hoiem

Remember…
• No classifier is inherently
better than any other: you
need to make assumptions to
generalize

• Three kinds of error

– Inherent: unavoidable
– Bias: due to over-simplifications
– Variance: due to inability to
perfectly estimate parameters
from limited data

Slide credit: D. Hoiem

• How to reduce variance?
– Choose a simpler classifier
– Regularize the parameters
– Get more training data
• How to reduce bias?
– Choose a more complex, more expressive classifier
– Remove regularization
– (These might not be safe to do unless you get more training data)

Slide credit: D. Hoiem

To be continued…