arXiv subjects

Ahmed Elgammal

Publications and source records attributed to Ahmed Elgammal.

58 records · Page 4Linked to original sources

arXiv2014

Collaborative Discriminant Locality Preserving Projections With its Application to Face Recognition

We present a novel Discriminant Locality Preserving Projections (DLPP) algorithm named Collaborative Discriminant Locality Preserving Projection (CDLPP). In our algorithm, the discriminating power of DLPP are further exploited from two aspects. On the one hand, the global optimum of class scattering is guaranteed via using the between-class scatter matrix to replace the original denominator of DLPP. On the other hand, motivated by collaborative representation, an $L_2$-norm constraint is imposed to the projections to discover the collaborations of dimensions in the sample space. We apply our algorithm to face recognition. Three popular face databases, namely AR, ORL and LFW-A, are employed for evaluating the performance of CDLPP. Extensive experimental results demonstrate that CDLPP significantly improves the discriminating power of DLPP and outperforms the state-of-the-arts.

cs.CV↗

arXiv2013

Visual-Semantic Scene Understanding by Sharing Labels in a Context Network

We consider the problem of naming objects in complex, natural scenes containing widely varying object appearance and subtly different names. Informed by cognitive research, we propose an approach based on sharing context based object hypotheses between visual and lexical spaces. To this end, we present the Visual Semantic Integration Model (VSIM) that represents object labels as entities shared between semantic and visual contexts and infers a new image by updating labels through context switching. At the core of VSIM is a semantic Pachinko Allocation Model and a visual nearest neighbor Latent Dirichlet Allocation Model. For inference, we derive an iterative Data Augmentation algorithm that pools the label probabilities and maximizes the joint label posterior of an image. Our model surpasses the performance of state-of-art methods in several visual tasks on the challenging SUN09 dataset.

cs.CV↗

arXiv2011

A Spectral Theory for Tensors

In this paper we propose a general spectral theory for tensors. Our proposed factorization decomposes a tensor into a product of orthogonal and scaling tensors. At the same time, our factorization yields an expansion of a tensor as a summation of outer products of lower order tensors . Our proposed factorization shows the relationship between the eigen-objects and the generalised characteristic polynomials. Our framework is based on a consistent multilinear algebra which explains how to generalise the notion of matrix hermicity, matrix transpose, and most importantly the notion of orthogonality. Our proposed factorization for a tensor in terms of lower order tensors can be recursively applied so as to naturally induces a spectral hierarchy for tensors.

math.SP↗

arXiv2011

Learning Hypergraph Labeling for Feature Matching

This study poses the feature correspondence problem as a hypergraph node labeling problem. Candidate feature matches and their subsets (usually of size larger than two) are considered to be the nodes and hyperedges of a hypergraph. A hypergraph labeling algorithm, which models the subset-wise interaction by an undirected graphical model, is applied to label the nodes (feature correspondences) as correct or incorrect. We describe a method to learn the cost function of this labeling algorithm from labeled examples using a graphical model training algorithm. The proposed feature matching algorithm is different from the most of the existing learning point matching methods in terms of the form of the objective function, the cost function to be learned and the optimization method applied to minimize it. The results on standard datasets demonstrate how learning over a hypergraph improves the matching performance over existing algorithms, notably one that also uses higher order information without learning.

cs.CV↗