S. Lawrence, C. L. Giles, A. C. Tsoi, and A. D. Back. Face recognition: A hybrid neural network approach. Technical Report UMIACS-TR-96-16 and CS-TR-3608, Institute for Advanced Computer Studies, University of Maryland, College Park, MD 2074.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....in stock index [109] as well as with fuzzy sets for function learning [110] and for real time tool condition monitoring [111] The following paragraphs show a classification among hybrid systems not belonging clearly to one of the presented hybrid classes. Hybrid Neural Network Systems. In [112], a hybrid neural network scheme for face recognition is proposed. The model combines local image sampling, a self organizing map neural network and a convolutional neural network. The system provides a measure of confidence in its output and classification error approaches zero when rejecting as ....

Lawrence S., Giles C.L., Tsoi A.C., Back A.D., Face Recognition: A Hybrid Neural Network Approach, Technical Report, UMIACS-TR-96-16 and CS-TR- 3608, Institute for Advanced Computer Studies, University of Maryland, College Park, MD 20742, 1996

....2.4 Algorithm 3: class constrained test train splits In some applications, the training set is constrained to contain a certain number of items from each class. For instance, the ORL face dataset [8] contains 10 images from each of 40 individuals. The face recognition experiments presented in [6, 9] examine accuracy by varying the size of the training set using 1, 3, or 5 images for each of the 40 classes. In general, we define k to be the number of items required from class k to be in the training set . This section shows how our technique may be applied to this problem, by extending ....

S. Lawrence, C. Giles, A. Tsoi, and A. Back. Face recognition: A hybrid neural network approach. Technical Report UMIACS-TR-96-16, University of Maryland, 1996.

....number of items from each class (typically to reflect the proportions of the classes in S) this is termed stratification (Kohavi, 1995) For instance, the ORL face dataset (Samaria Harter, 1994) contains 10 images from each of 40 individuals. The face recognition experiments presented in (Lawrence et al. 1996; Sim et al. 2000) examine accuracy by varying the size of the training set using 1, 3, or 5 images for each of the 40 classes. In general, we define c to be the number of items required from class c to be in the training set. This section shows how complete cross validation may be applied to ....

Lawrence, S., Giles, C., Tsoi, A., & Back, A. (1996). Face recognition: A hybrid neural network approach (Technical Report UMIACS-TR-96-16). University of Maryland.

....number of training images (n) Experiments were conducted on the FERET database with the original images subsampled to 16 16. clustered in the weight space, so the actual training data is not needed. In PCA 2, a memory based variant of PCA, each of the weight vectors is individually stored [13] requiring more storage space, but providing PCA 2 with a richer representation. When a test image is presented to the system, it is first projected into the eigenspace (by Equation 1) and its weight vector new is computed. new is then compared against the stored weight vectors, ....

....4 which achieved an accuracy result of 94.8 in its best run, was significantly outperformed (in identical experiments) by ARENA (L 0 without synthetic images: 96.2 , with synthetic images, 97.1 ) 6.2. Comparisons with other algorithms We have also duplicated some experiments reported in [13]. They examined the performance of four algorithms, Eigenfaces average per class (identical to PCA 1) Eigenfaces one per image (identical to PCA 2) PCA CN (PCA combined with a convolutional network classifier) and SOM CN (Self Organizing Map combined with a CN) on the ORL dataset ....

[Article contains additional citation context not shown here]

S. Lawrence, C. Giles, A. Tsoi, and A. Back. Face recognition: A hybrid neural network approach. Technical Report UMIACS-TR-96-16, University of Maryland, 1996.

....2.4 Algorithm 3: class constrained test train splits In some applications, the training set is constrained to contain a certain number of items from each class. For instance, the ORL face dataset [8] contains 10 images from each of 40 individuals. The face recognition experiments presented in [6, 9] examine accuracy by varying the size of the training set using 1, 3, or 5 images for each of the 40 classes. In general, we define k to be the number of items required from class k to be in the training set 8 . This section shows how our technique may be applied to this problem, by extending ....

S. Lawrence, C. Giles, A. Tsoi, and A. Back. Face recognition: A hybrid neural network approach. Technical Report UMIACS-TR-96-16, University of Maryland, 1996.

....for each person s images in the training set is computed and stored [24] PCA 1 assumes that each person s face images will be clustered in the weight space, so the actual training data is not needed. In PCA 2, a memory based variant of PCA, each of the weight vectors is individually stored [12] requiring more storage space, but providing PCA 2 with a richer representation. When a test image is presented to the system, it is first projected into the eigenspace (by Equation 1) and its weight vector new is computed. new is then compared against the stored weight vectors, ....

.... Images per person 1 3 5 Eigenface avg per class 61.4 71.1 74.0 Eigenface one per img 61.4 81.8 89.5 PCA CN 65.8 76.8 92.5 SOM CN 70.0 88.2 96.5 ARENA (p = 0,s = 16) 74.7 92.2 97.1 ARENA (p = 1,s = 16) 75.1 92.0 96.8 Table 1: Comparison of ARENA with results reported in [12]. 7.2 Comparisons with other algorithms We have also duplicated a set of experiments reported in [12] They examined the performance of four algorithms, Eigenfaces average per class (identical to PCA 1) Eigenfaces one per image (identical to PCA 2) PCA CN (PCA combined with a ....

[Article contains additional citation context not shown here]

S. Lawrence, C. Giles, A. Tsoi, and A. Back. Face recognition: A hybrid neural network approach. Technical Report UMIACS-TR-96-16, University of Maryland, 1996.

....and Face Recognition A lot of papers have been written about applying NNs to face recognition. Research in this area can be divided into two main categories: 1: using NNs for face recognition, when a whole picture of a face is considered as a pattern and the objective is to identify the face [4], and 2: using NNs for locating facial features, where a facial image is considered as a set of patterns, each of which contains information about some part of the face, and the objective is to extract this information, i.e. to recognize those patterns that contain a feature of interest. In this ....

....association of pixel values in an input layer and an answer ( eye , non eye ) in the output layer, which consists of a single neuron . This is also the case with an ALN. Some other types of NNs have also been applied to eyes: Radial Basis Function networks [9] and Self Organizing Maps (SOM) [5, 4]. Following the order of questions in Section 3.1, let us describe the observations and solutions proposed by other researchers. First, experiments have been carried out with 256x256, 128x128 and 64x64 images, and it has been found [7] that processing of low resolution images is more robust than ....

[Article contains additional citation context not shown here]

S. Lawrence, C. L. Giles, A. C. Tsoi, A. D. Back. Face Recognition: A Hybrid Neural Network Approach, IEEE Trans. on Neural Networks, special issue on Pattern Recognition, accepted for

No context found.

S. Lawrence, C. L. Giles, A. C. Tsoi, and A. D. Back. Face recognition: A hybrid neural network approach. Technical Report UMIACS-TR-96-16 and CS-TR-3608, Institute for Advanced Computer Studies, University of Maryland, College Park, MD 2074.

No context found.

Lawrence, S., Giles, C. L., Chung Tsoi, A., Back, A. D.: Face Recognition: A Hybrid Neural Network Approach. Technical Report, UMIACS-TR-96-16 and CS-TR-3608. Institute for Advanced Computer Studies University of Maryland, College Park (1996).

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