C. Liu and H. Wechsler. Evolutionary pursuit and its application to face recognition. IEEE Transactions on Pattern Analysis and Machine Intelligence, 22(6):570-- 582, 2000.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....Then, the main drawback of FLD is that could be over adjusted on the training images, and then the recognition system may have an important lack of generalization that may be notice in the recognition rate. 2.2.3. Evolutionary Pursuit EP The eigenspace based adaptive approach proposed in [1], searches for the best set of projection axes in order to maximize a fitness function, measuring at the same time the classification accuracy and generalization ability of the system. Because the dimension of the solution space of this problem is too large, it is solved using a specific kind of ....

....that determines the importance of the second term against the first one. The generalization ability is computed as: z s (a k , a i ) m (i ) i =1 . 12) where m is the global mean and m is the mean of the corresponding classC i . The accuracy measure z a (a k , a i ) proposed in [1] is just the recognition rate of training face images as the top choice. But in our implementation we realize that this measure became always 100 when the number of classes is small respect to the dimension of the reduced space (we use only 15 classes against the 369 used in [1] Then we use the ....

[Article contains additional citation context not shown here]

C. Liu and H. Wechsler, "Evolutionary Pursuit and Its Application to Face Recognition", IEEE Trans. Patt. Analysis and Machine Intell., vol. 22, no. 6, 570-582, 2000.

....the reduced representation of each database image. These representations are the ones to be used in the recognition process. Among the projection methods employed for the reduction of dimensionality, we can mention: PCA [10] Linear Discriminant Analysis (LDA) 2] and Evolutionary Pursuit (EP) [3]. Among the similarity matching criteria employed for the recognition process, they have been used: Euclidean , Cosine and Mahalanobis distance, Self Organizing Map (SOM) clustering, and Fuzzy Feature Contrast (FFC) similarity (see definitions in [6] All this methods have been analyzed and ....

Liu C., and Wechsler H., "Evolutionary Pursuit and Its Application to Face Recognition", IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 22, no. 6, pp. 570-582, June 2000.

....main drawback of FLD is that it could be over adjusted on the target images, and then the recognition system may have an important lack of generalization that may be notice in the resulting system s recognition rate. Evolutionary Pursuit EP The eigenspace based adaptive approach proposed in [4], searches for the best set of projection axes in order to maximize a fitness function, measuring at the same time the classification accuracy and generalization ability of the system. Because the dimension of the solution space of this problem is too large, it is solved using a specific kind of ....

....term against the first one. The generalization ability is computed as (remember b S given by (6) s ( k , a i ) m (i ) i = 1 . 12) where m is the global mean and m is the mean of the corresponding class C i . Although the accuracy measure a ( k , a i ) proposed in [4] is the recognition rate of training face images as the top choice, in our implementation we used the top 2 identity because with the top choice it was too easy to obtain a 100 recognition rate. As usual, in order to find the maximal value of the fitness function, a random set of chromosomes is ....

[Article contains additional citation context not shown here]

Liu C., and Wechsler H., "Evolutionary Pursuit and Its Application to Face Recognition", IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 22, no. 6, pp. 570-582, June 2000.

....associated with the largest general eigenvalues (Fisher Parameters [4] To solve the problem of the large size of the scatter matrices, PCA is applied before FLD. In this way we are also solving the problem of singularity for # w . Evolutionary Pursuit EP EP, originally proposed in [3], searches for the best set of projection axes in order to maximize a fitness function that measures, at the same time, the classification accuracy and generalization ability of the system. Because the dimension of the solution space of this problem is very large, it is solved using Genetic ....

....a certain projection system. To evaluate this system the following fitness function is used: za### ki a = za lza aki ski aa ### ### , 2) where za aki a ### measures the accuracy, za ski a ### measures the generalization ability, and l is a positive constant (see definitions in [3]) 2.3 Similarity Matching Methods Euclidean Distance d### # ## # ## #### = 3) Cosine Distance #### # # ## ## . 4) Mahalanobis Distance d##### # # # # # = ## ; #: correlation matrix. 5) From a geometrical point of view this distance has a scaling effect in ....

# Liu C., and Wechsler H., Evolutionary Pursuit and Its Application to Face Recognition, IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 22, no. 6, pp. 570-582, June 2000.

No context found.

C. Liu and H. Wechsler, "Evolutionary pursuit and its application to face recognition," IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 22, no. 6, pp. 570--582, 2000. 20

No context found.

C. Liu and H. Wechsler, "Evolutionary pursuit and its application to face recognition," IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 22, no. 6, pp. 570--582, 2000.

No context found.

C. Liu and H. Wechsler, "Evolutionary pursuit and its application to face recognition," IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 22, no. 6, pp. 570--582, 2000.

No context found.

C. Liu and H. Wechsler, "Evolutionary pursuit and its application to face recognition," IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 22, no. 6, pp. 570--582, 2000.

No context found.

C. Liu and H. Wechsler, "Evolutionary pursuit and its application to face recognition," IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 22, no. 6, pp. 570--582, 2000. 20

No context found.

C. Liu and H. Wechsler, "Evolutionary pursuit and its application to face recognition," IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 22, no. 6, pp. 570--582, 2000.

....[13] 7] or a pooled within class covariance matrix [10] In particular, EFC achieves 98.5 recognition rate using only 25 features. 2. Background Learning to recognize visual objects, such as human faces, requires the ability to find meaningful patterns in spaces of very high dimensionality [16]. Psychophysical findings indicate, however, that perceptual tasks such as similarity judgment tend to be performed on a low dimensional representation of the sensory data. Low dimensionality is especially important for learning, as the number of examples required for attaining a given level of ....

....driven coding schemes are optimal and useful only with respect to data compression and decorrelation of low (2nd) order statistics. The recognition aspect is not considered and one should thus not expect optimal performance for tasks such as face recognition when using PCAlike coding schemes [14] [16]. The Fisher Linear Discriminant (FLD) is a popular discriminant method for the very purpose of achieving high separability between the different patterns in whose classification one is interested. Characteristic of this approach are recent but similar methods such as the Most Discriminating ....

C. Liu and H. Wechsler, "Evolutionary pursuit and its application to face recognition," IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 22, no. 6, pp. 570--582, 2000.

No context found.

C. Liu and H. Wechsler. Evolutionary pursuit and its application to face recognition. IEEE Transactions on Pattern Analysis and Machine Intelligence, 22(6):570-- 582, 2000.

No context found.

C. Liu and H. Wechsler. Evolutionary pursuit and its application to face recognition. IEEE Transactions on Pattern Analysis and Machine Intelligence, 22(6):570--582, June 2000.

No context found.

C. Liu and H. Wechsler. Evolutionary pursuit and its application to face recognition. IEEE Trans. on Pattern Analysis and Machine Intelligence, 22:570--582, June 2000.

No context found.

C. Liu and H. Wechsler. Evolutionary pursuit and its application to face recognition. IEEE Transactions on Pattern Analysis and Machine Intelligence, 22(6):570-- 582, 2000.

No context found.

C. Liu and H. Wechsler, "Evolutionary pursuit and its application to face recognition," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 22, no.6, pp. 570--582, 2000.

No context found.

C. Liu and H. Wechsler. Evolutionary pursuit and its application to face recognition. IEEE Trans. on Pattern Analysis and Machine intelligence, 22(6), 2000.

No context found.

C. Liu and H. Wechsler, "Evolutionary pursuit and its application to face recognition," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 22, no.6, pp. 570--582, 2000.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC