Method for boosting the performance of machine-learning classifiers

Claim 2

A computer-implemented process for using feature selection to obtain a strong classifier from a combination of weak classifiers, said process comprising: (a) an inputting step for inputting a set of training examples, a prescribed maximum number of weak classifiers, a cost function capable of measuring the overall cost, and an acceptable maximum cost; (b) a weak classifier computing set for computing a set of weak classifiers, each classifier being associated to a particular feature of the training examples, (c) a most significant weak classifier determining step for determining which of the set of weak classifiers is the most significant classifier; (d) an adding step for adding said most significant classifier to a current set of optimal weak classifiers; (e) a least significant classifier determining step for determining which of the current set of optimal weak classifiers is the least significant classifier; (f) a first cost computing step for computing the overall cost for the current set of optimal weak classifiers using the cost function; (g) a least significant classifier conditional removing step for conditionally removing the least significant classifier for the current set of optimal weak classifiers; (h) an overall cost computing step for computing the overall cost for the current set of optimal weak classifiers less the least significant classifier using the cost function; (i) a lower overall cost computing step for determining whether the removal of the least significant classifier results in a lower overall cost; (j) whenever it is determined that the removal of the least significant classifier results in a lower overall cost, a least significant classifier eliminating step for eliminating the least significant classifier; (k) a recomputation step for recomputing each classifier in the current set of optimal weak classifiers associated with a feature added subsequent to the eliminated classifier while keeping the earlier optimal weak classifiers unchanged; (l) repeating steps (f) through (k) until it is determined the removal of the least significant classifier does not result in a lower overall cost and then reinstating the last identified least significant classifier to the current set of optimal weak classifiers; (m) a step for determining if the number of weak classifiers in the current set of optimal weak classifiers equals the prescribed maximum number of weak classifiers or the last computed overall cost for the current set of optimal weak classifiers is less than the acceptable maximum cost; and (n) whenever it is determined that the number of weak classifiers in the current set of optimal weak classifiers does not equal the prescribed maximum number of weak classifiers and the last computed overall cost for the current set of optimal weak classifiers exceeds the acceptable maximum cost, repeating steps (c) through (m) until it is determined that the number of weak classifiers in the current set of optimal weak classifiers does equal the prescribed maximum number of weak classifiers or the last computed overall cost for the current set of optimal weak classifiers becomes less than the maximum allowable cost, then outputting the sum of the individual weak classifiers as the trained strong classifier.

Claim 3

The computer-implemented process of claim 1 wherein the weak classifier determining set comprises a step for deriving each classifier based on a histogram of a scalar value feature for face training examples and a histogram of a scalar value feature for non-face training examples.

Claim 4

The computer-implemented process of claim 1 wherein the most significant classifier includes the feature that is the most likely to predict whether a training example matches the classification of a particular classifier.

Claim 5

The computer-implemented process of claim 1 wherein the set of weak classifiers are designed to classify whether a training example is a face or non-face.

Claim 6

The computer-implemented process of claim 1 wherein the set of weak classifiers is designed to classify a training example as a text type.

Claim 7

The computer-implemented process of claim 1 wherein the set of weak classifiers is designed to classify a training example as a type of document.

Claim 8

The computer-implemented process of claim 1 wherein the set of weak classifiers is designed to classify a training example as a speech pattern.

Claim 9

The computer-implemented process of claim 1 wherein the set of weak classifiers is designed to classify a training example as a type of medical condition.

Claim 10

The process of claim 1 wherein a weak classifier h* j (x) is computed as h j * ⁡ ( x ) ⩵ 1 2 ⁡ [ log ⁢ P j ⁡ ( x ❘ y = + 1 , w ) P j ⁡ ( x ❘ y = - 1 , w ) + log ⁢ P ⁡ ( y = + 1 ) P ⁡ ( y = - 1 ) ] wherein the probability densities of a feature j for a sub-sample x of a training example is denoted by P j (x|y=+1) for a sought pattern and P j (x|y=−1) for a non-sought pattern and the normalized weights are denoted by w.

Claim 11

The computer-implemented process of claim 9 wherein the probability density for a sought pattern and the probability density for a non-sought pattern can be estimated using the histograms resulting from weighted voting of the training examples.

Claim 12

The computer-implemented process of claim 9 wherein the step for determining which of the set of weak classifiers is the most significant classifier comprises a step for defining the most significant classifier h M (x) as, h M ⁡ ( x ) = a ⁢ rg ⁢ min h * ⁢ ε ⁢ ⁢ H w * ⁢ ∑ ⁢ ⅇ - y i ⁡ [ h ⁡ ( x i ) + h * ⁡ ( x i ) ] , wherein H w * = { h j * ⁡ ( x ) ❘ ∀ j } , h ⁡ ( x ) = ∑ m = 1 M - 1 ⁢ ⁢ h m ⁡ ( x ) , and M is the total number of weak classifiers in the set of weak classifiers.

Claim 13

The computer-implemented process of claim 9 wherein the step for determining which of the set of weak classifiers is the least significant classifier comprises defining the least significant classifier h′(x) as, h′=arg min hεH W J(H M −h) where H M denotes the strong classifier built upon the current set H M of selected weak classifiers.

Claim 14

The computer-implemented process of claim 1 wherein the step for computing the overall cost comprises computing the overall cost J(h(x)) as J ⁡ ( h ⁡ ( x ) ) = ∑ i ⁢ ⁢ ⅇ - y i ⁢ h ⁡ ( x i ) wherein y=+1 for a sought pattern and y=−1 for a nonsought pattern and h(x i ) is a weak classifier in the set of weak classifiers.

Claim 15

The computer-implemented process of claim 1 wherein outputting the sum of the individual weak classifiers as the trained strong classifier comprises outputting the sum H ⁡ ( x ) ⁢ ⁢ as ⁢ ⁢ H ⁡ ( x ) = sign [ ⁢ ∑ m = 1 M ⁢ ⁢ h m ⁡ ( x ) ] wherein M is the total number of weak classifiers in the set of weak classifiers h m (x) is a weak classifier in the current set of weak classifiers.

Claim 16

A computer-implemented process for detecting a person's face in an input image and identifying a face pose range into which the face pose exhibited by the detected face falls comprises: a database computing step for creating a database comprising a plurality of training feature characterizations, each of which characterizes the face of a person at a known face pose or a non-face; a training step for training a plurality of detectors arranged in a pyramidal architecture to determine whether a portion of an input image depicts a person's face having a face pose falling within a face pose range associated with one of the detectors using the training feature characterizations; and wherein said detectors using a greater number of feature characterizations are arranged at the bottom of the pyramid, and said detectors arranged to detect finer ranges of face pose are arranged at the bottom of the pyramid; and wherein the program module to train a plurality of detectors comprises sub-modules to, (a) an inputting step for inputting a set of training examples, a prescribed maximum number of weak classifiers, a cost function capable of measuring the overall cost, and an acceptable maximum cost; (b) a weak classifier computing step for computing a set of weak classifiers, each classifier being associated to a particular feature of the training examples, (c) a most significant classifier determining step for determining which of the set of weak classifiers is the most significant classifier; (d) an adding step for adding said most significant classifier to a current set of optimal weak classifiers; (e) a least significant classifier determining for determining which of the current set of optimal weak classifiers is the least significant classifier; (f) a first cost computing step for computing the overall cost for the current set of optimal weak classifiers using the cost function; (g) a least significant classifier conditional removing step for conditionally removing the least significant classifier for the current set of optimal weak classifiers; (h) an overall cost computing step for computing the overall cost for the current set of optimal weak classifiers less the least significant classifier using the cost function; (i) a lower overall cost determining step for determining whether the removal of the least significant classifier results in a lower overall cost; (j) whenever it is determined that the removal of the least significant classifier results in a lower overall cost, an elimination step for eliminating the least significant classifier; (k) a recomputation step for recomputing each classifier in the current set of optimal weak classifiers associated with a feature added subsequent to the eliminated classifier while keeping the earlier optimal weak classifiers unchanged; (l) repeating steps (f) through (k) until it is determined the removal of the least significant classifier does not result in a lower overall cost and then reinstate the last identified least significant classifier to the current set of optimal weak classifiers; (m) a step for determining if the number of weak classifiers in the current set of optimal weak classifiers equals the prescribed maximum number of weak classifiers or the last computed overall cost for the current set of optimal weak classifiers is less than the acceptable maximum cost; and (n) whenever it is determined that the number of weak classifiers in the current set of optimal weak classifiers does not equal the prescribed maximum number of weak classifiers and the last computed overall cost for the current set of optimal weak classifiers exceeds the acceptable maximum cost, repeating steps (c) through (m) until it is determined that the number of weak classifiers in the current set of optimal weak classifiers does equal the prescribed maximum number of weak classifiers or the last computed overall cost for the current set of optimal weak classifiers becomes less than the maximum allowable cost, then outputting the sum of the individual weak classifiers as the trained strong classifier.

Claim 17

A computer-implemented process for detecting a person's face in an input image and identifying a face pose range into which the face pose exhibited by the detected face falls comprises: a database computing step for creating a database comprising a plurality of training feature characterizations, each of which characterizes the face of a person at a known face pose or a non-face; a training step for training a plurality of detectors arranged in a pyramidal architecture to determine whether a portion of an input image depicts a person's face having a face pose falling within a face pose range associated with one of the detectors using the training feature characterizations; and wherein said detectors using a greater number of feature characterizations are arranged at the bottom of the pyramid, and said detectors arranged to detect finer ranges of face pose are arranged at the bottom of the pyramid; and wherein the program module to train a plurality of detectors comprises sub-modules to, (a) an inputting step for inputting a set of training examples, a prescribed maximum number of weak classifiers, a cost function capable of measuring the overall cost, and an acceptable maximum cost; (b) a weak classifier computing step for computing a set of weak classifiers, each classifier being associated to a particular feature of the training examples, (c) a most significant classifier determining step for determining which of the set of weak classifiers is the most significant classifier; (d) an adding step for adding said most significant classifier to a current set of optimal weak classifiers; (e) a least significant classifier determining for determining which of the current set of optimal weak classifiers is the least significant classifier; (f) a first cost computing step for computing the overall cost for the current set of optimal weak classifiers using the cost function; (g) a least significant classifier conditional removing step for conditionally removing the least significant classifier for the current set of optimal weak classifiers; (h) an overall cost computing step for computing the overall cost for the current set of optimal weak classifiers less the least significant classifier using the cost function; (i) a lower overall cost determining step for determining whether the removal of the least significant classifier results in a lower overall cost; (j) whenever it is determined that the removal of the least significant classifier results in a lower overall cost, an elimination step for eliminating the least significant classifier; (k) a recomputation step for recomputing each classifier in the current set of optimal weak classifiers associated with a feature added subsequent to the eliminated classifier while keeping the earlier optimal weak classifiers unchanged; (l) repeating steps (f) through (k) until it is determined the removal of the least significant classifier does not result in a lower overall cost and then reinstate the last identified least significant classifier to the current set of optimal weak classifiers; (m) a step for determining if the number of weak classifiers in the current set of optimal weak classifiers equals the prescribed maximum number of weak classifiers or the last computed overall cost for the current set of optimal weak classifiers is less than the acceptable maximum cost; and (n) whenever it is determined that the number of weak classifiers in the current set of optimal weak classifiers does not equal the prescribed maximum number of weak classifiers and the last computed overall cost for the current set of optimal weak classifiers exceeds the acceptable maximum cost, repeating steps (c) through (m) until it is determined that the number of weak classifiers in the current set of optimal weak classifiers does equal the prescribed maximum number of weak classifiers or the last computed overall cost for the current set of optimal weak classifiers becomes less than the maximum allowable cost, then outputting the sum of the individual weak classifiers as the trained strong classifier.