Genetic Feature Subset Selection for Gender Classification: A Comparison Study

1
Genetic Feature Subset Selection for Gender
Classification A Comparison Study

• Zehang Sun, George Bebis, Xiaojing Yuan, and
  Sushil Louis
• Computer Vision Laboratory
• Department of Computer Science
• University of Nevada, Reno
• bebis_at_cs.unr.edu
• http//www.cs.unr.edu/CVL

2
Gender Classification

• Problem statement
• Determine the gender of a subject from facial
  images.
• Potential applications
• Face Recognition
• Human-Computer Interaction (HCI)
• Challenges
• Race, age, facial expression, hair style, etc.

3
Gender Classification by Humans

• Humans are able to make fast and accurate gender
  classifications.
• It takes 600 ms on the average to classify faces
  according to their gender (Bruce et al.,1987).
• 96 accuracy has been reported using photos of
  non-familiar faces without hair information
  (Bruce et. al., 1993).
• Empirical evidence indicates that gender
  decisions are always made much faster than
  identity.
• Computation of gender and identity might be two
  independent processes.
• There is evidence that gender classification is
  carried out by a separate population of cells in
  the inferior temporal cortex (Damasio et. al.,
  1990).

4
Designing a Gender Classifier

• The majority of gender classification schemes are
  based on supervised learning.
• Definition
• Feature extraction determines an appropriate
  subspace of dimensionality m in the original
  feature space of dimensionality d (m ltlt d).

5
Previous Approaches

• Geometry-based
• Use distances, angles, and areas among facial
  features.
• Point-to-point distances discriminant analysis
  (Burton 93, Fellous 97)
• Feature-to-feature distances HyberBF NNs
  (Brunelli 92)
• Wavelet features elastic graph matching
  (Wiskott 95)
• Appearance-based
• Raw images NNs (Cottrell 90, Golomb 91, Yen
  94)
• PCA NNs (Abdi 95),
• PCA nearest neighbor (Valentin 97)
• Raw images SVMs (Moghaddam 02)

6
What Information is Useful for Gender
Classification?

• Geometry-based approaches
• Representing faces as a set of features assumes
  a-priori knowledge about what are the features
  and/or what are the relationships between them.
• There is no simple set of features that can
  predict the gender of faces accurately.
• There is no simple algorithm for extracting the
  features automatically from images.
• Appearance-based approaches
• Certain features are nearly characteristic of one
  sex or the other (e.g., facial hair for men,
  makeup or certain hairstyles for women).
• Easier to represent this kind of information
  using appearance-based feature extraction
  methods.
• Appearance-based features, however, are more
  likely to suffer from redundant and irrelevant
  information.

7
Feature Extraction Using PCA

• Feature extraction is performed by projecting the
  data in a lower-dimensional space using PCA.
• PCA maps the data in a lower-dimensional space
  using a linear transformation.
• The columns of the projection matrix are the
  best eigenvectors (i.e., eigenfaces) of the
  covariance matrix of the data.

8
Which Eigenvectors Encode Mostly Gender-Related
Information?
Sometimes, it is possible to determine what
features are encoded by specific eigenvectors.
9
Which Eigenvectors Encode Mostly Gender-Related
Information? (contd)

• All eigenvectors contain information relative to
  the gender of faces, however, only the
  information conveyed by eigenvectors with large
  eigenvalues can be generalized to new faces (Abdi
  et al, 1995).
• Removing specific eigenvectors could in fact
  improve performance (Yambor et al, 2000)

10
Critique of Previous Approaches

• No explicit feature selection is performed.
• Same features used for face identification are
  also used for gender classification.
• Some features might be redundant or irrelevant.
• Rely heavily on the classifier.
• Classification accuracy can suffer.
• Time consuming training and classification.

11
Project Goal

• Improve the performance of gender classification
  using feature subset selection.

12
Feature Selection
What constitutes a good set of features for
classification?

• Definition
• Given a set of d features, select a subset of
  size m that leads to the smallest classification
  error.
• Filter Methods
• Preprocessing steps performed independent of the
  classification algorithm or its error criteria.
• Wrapper Methods
• Search through the space of feature subsets using
  the criterion of the classification algorithm to
  select the optimal feature subset.
• Provide more accurate solutions than filter
  methods, but in general are more computationally
  expensive.

13
What are the Benefits?

• Eliminate redundant and irrelevant features.
• Less training examples are required.
• Faster and more accurate classification.

14
Project Objectives

• Perform feature extraction by projecting the
  images in a lower-dimensional space using
  Principal Components Analysis (PCA).
• Perform feature selection in PCA space using
  Genetic Algorithms.
• Test four traditional classifiers (Bayesian, LDA,
  NNs, and SVMs).
• Compare with traditional feature subset selection
  approaches (e.g., Sequential Backward Floating
  Search (SBFS)).

15
Genetic Algorithms (GAs) Review

• What is a GA?
• An optimization technique for searching very
  large spaces.
• Inspired by the biological mechanisms of natural
  selection and reproduction.
• What are the main characteristics of a GA?
• Global optimization technique.
• Uses objective function information, not
  derivatives.
• Searches probabilistically using a population of
  structures (i.e., candidate solutions using some
  encoding).
• Structures are modified at each iteration using
  selection, crossover, and mutation.

16
Structure of GA

• 10010110 10010110
• 01100010 01100010
• 10100100... 10100100
• 10010010 01111001
• 01111101 10011101

Evaluation and Selection
Crossover
Mutation
Current Generation
Next Genaration
17
Encoding and Fitness Evaluation

• Encoding scheme
• Transforms solutions in parameter space into
  finite length strings (chromosomes) over some
  finite set of symbols.
• Fitness function
• Evaluates the goodness of a solution.

18
Selection Operator

• Probabilistically filters out solutions that
  perform poorly, choosing high performance
  solutions to exploit.
• Chromosomes with high fitness are copied over to
  the next generation.

fitness
19
Crossover and Mutation Operators

• Generate new solutions for exploration.
• Crossover
• Allows information exchange between points.
• Mutation
• Its role is to restore lost genetic material.

Mutated bit
20
Genetic Feature Subset Selection

• Binary encoding
• Fitness evaluation

EV1
EV250
(search using first 250 eigenvectors)
fitness104?accuracy 0.4 ? zeros
accuracy from validation set
number of features
21
Genetic Feature Subset Selection (contd)

• Cross-generational selection strategy
• Assuming a population of size N, the offspring
  double the size of the population, and we select
  the best N individuals from the combined
  parent-offspring population.
• GA parameters
• Population size 350
• Number of generations 400
• Crossover rate 0.66
• Mutation rate 0.04

22
Dataset

• 400 frontal images from 400 different people
• 200 male, 200 female
• Different races
• Different lighting conditions
• Different facial expressions
• Images were registered and normalized
• No hair information
• Account for different lighting conditions

23
Experiments

• Gender classifiers
• Linear Discriminant Analysis (LDA)
• Bayes classifier
• Neural Network (NN) classifier
• Support Vector Machine (SVM) classifier
• Three - fold cross validation
• Training set 75 of the data
• Validation set 12.5 of the data
• Test set 12.5 of the data

24
Classification Error Rates
22.4
17.7
14.2
13.3
11.3
8.9
9
6.7
4.7
ERM error rate using manually selected feature
subsets ERG error rate using GA selected
feature subsets
25
Ratio of Features - Information Kept
69
61.2
42.8
38
36.4
32.4
31
17.6
13.3
8.4
RN percentage of number of features in the
feature subset RI percentage of information
contained in the feature subset.
26
Distribution of Selected Eigenvectors
(a) LDA
(b) Bayes
(d) SVMs
(c) NN
27
Reconstructed Images
Original images
Using top 30 EVs
Using EVs selected by LDA-PCAGA
Using EVs selected by B-PCAGA
28
Reconstructed Images (contd)
Original images
Using top 30 EVs
Using EVs selected by NN-PCAGA
Using EVs selected by SVM-PCAGA
Reconstructed faces using GA-selected EVs have
lost information about identity but do disclose
strong gender information
Certain gender-irrelevant features do not appear
in the reconstructed images using GA-selected
EVs
29
Comparison with SBFS

• Sequential Backward Floating Search (SBFS) is a
  combination of two heuristic search schemes
• (1) Sequential Forward Selection (SFS)
• - starts with an empty feature set and at each
  set selects the best single feature to be added
  to the feature subset.
• (2) Sequential Backward Selection (SBS).
• - starts with the entire feature and at each step
  drops the feature whose absence least decreases
  the performance.

30
Comparison with SBFS (contd)

• SBFS is an advanced version of plus l - take away
  r method that first enlarges the feature subset
  by l features using forward selection and then
  removes r features using backward selection.
• The number of forward and backward steps in SBFS
  is dynamically controlled and updated based on
  the classifiers performance.

31
Comparison with SBFS (contd)
(b) SVMsGA
(a) SVMsSBFS
ERM error rate using the manually selected
feature subsets ERG error rate using GA
selected feature subsets. ERSBFS error rate
using SBFS
32
Comparison with SBFS (contd)
Original images
Using top 30 EVs
Using EVs selected by SVM-PCAGA
Using EVs selected by SVM-PCASBFS
33
Conclusions

• We have considered the problem of gender
  classification from frontal facial images using
  genetic feature subset selection.
• GAs provide a simple, general, and powerful
  framework for feature subset selection.
• Very useful, especially when the number of
  training examples is small.
• We have tested four well-known classifiers using
  PCA for feature extraction.
• Genetic subset feature selection has led to lower
  error rates in all cases.

34
Future Work

• Generalize feature encoding scheme.
• Use weights instead of 0/1 encoding.
• Consider more powerful fitness functions.
• Use larger data sets.
• FERET data set.
• Apply feature selection using different features.
• Various features (e.g., Wavelet or Gabor
  features)
• Experiment with different data sets.
• Different data sets (e.g., vehicle detection)