221 results found
Zafeiriou S, Tefas A, Pitas I, 2007, Learning Discriminant Person-Specific Facial Models Using Expandable Graphs, IEEE Transactions on Information Forensics and Security, Vol: 2, Pages: 55-68, ISSN: 1556-6013
Goudelis G, Zafeiriou S, Tefas A, et al., 2007, A Novel Kernel Discriminant Analysis for Face Verification., Publisher: IEEE, Pages: 493-496
Zafeiriou S, Tefas A, Pitas I, 2007, Discriminant Graph Structures for Face Verification., Publisher: IEEE, Pages: 497-500
Nikolopoulos S, Zafeiriou S, Sidiropoulos P, et al., 2006, Image Replica Detection using R-Trees and Linear Discriminant Analysis, 2006 IEEE International Conference on Multimedia and Expo, Publisher: IEEE
Zafeiriou S, Tefas A, Buciu I, et al., 2006, Exploiting discriminant information in nonnegative matrix factorization with application to frontal face verification, IEEE Transactions on Neural Networks, Vol: 17, Pages: 683-695, ISSN: 1045-9227
Zafeiriou S, Tefas A, Pitas I, 2005, Blind Robust Watermarking Schemes for Copyright Protection of 3D Mesh Objects, IEEE Transactions on Visualization and Computer Graphics, Vol: 11, Pages: 596-607, ISSN: 1077-2626
Zafeiriou S, Tefas A, Pitas I, 2005, Exploiting discriminant information in elastic graph matching., Publisher: IEEE, Pages: 768-771
Zafeiriou S, Tefas A, Buciu I, et al., 2005, Class-Specific Discriminant Non-negative Matrix Factorization for Frontal Face Verification., Publisher: Springer, Pages: 206-215
Zafeiriou S, Tefas A, Pitas I, 2004, A blind robust watermarking scheme for copyright protection of 3d mesh models., Publisher: IEEE, Pages: 1569-1572
Chrysos GG, Panagakis Y, Zafeiriou S, Visual Data Augmentation through Learning
The rapid progress in machine learning methods has been empowered by i) hugedatasets that have been collected and annotated, ii) improved engineering (e.g.data pre-processing/normalization). The existing datasets typically includeseveral million samples, which constitutes their extension a colossal task. Inaddition, the state-of-the-art data-driven methods demand a vast amount ofdata, hence a standard engineering trick employed is artificial dataaugmentation for instance by adding into the data cropped and (affinely)transformed images. However, this approach does not correspond to any change inthe natural 3D scene. We propose instead to perform data augmentation through learning realisticlocal transformations. We learn a forward and an inverse transformation thatmaps an image from the high-dimensional space of pixel intensities to a latentspace which varies (approximately) linearly with the latent space of arealistically transformed version of the image. Such transformed images can beconsidered two successive frames in a video. Next, we utilize thesetransformations to learn a linear model that modifies the latent spaces andthen use the inverse transformation to synthesize a new image. We argue thatthe this procedure produces powerful invariant representations. We perform bothqualitative and quantitative experiments that demonstrate our proposed methodcreates new realistic images.
Zafeiriou L, Nicolaou MA, Zafeiriou S, et al., Learning Slow Features for Behaviour Analysis
This data is extracted from the Web of Science and reproduced under a licence from Thomson Reuters. You may not copy or re-distribute this data in whole or in part without the written consent of the Science business of Thomson Reuters.