Locality Constraint Dictionary Learning With Support Vector for Pattern Classification

Abstract

Discriminative dictionary learning (DDL) has recently gained significant attention due to its impressive performance in various pattern classification tasks. However, the locality of atoms is not fully explored in conventional DDL approaches which hampers their classification performance. In this paper, we propose a locality constraint dictionary learning with support vector discriminative term (LCDL-SV), in which the locality information is preserved by employing the graph Laplacian matrix of the learned dictionary. To jointly learn a classifier during the training phase, a support vector discriminative term is incorporated into the proposed objective function. Moreover, in the classification stage, the identity of test data is jointly determined by the regularized residual and the learned multi-class support vector machine. Finally, the resulting optimization problem is solved by utilizing the alternative strategy. Experimental results on benchmark databases demonstrate the superiority of our proposed method over previous dictionary learning approaches on both hand-crafted and deep features.

CSI Amplitude Fingerprinting for Indoor Localization with Dictionary Learning

“…In the first layer, we introduced an incoherent promotion term to ensure the sparse representation ability of CSI data in the corresponding region sub-dictionary. In order to make sparse coding also have the ability to locate fingerprints, we first take the sparse coding of the first layer sub-dictionary as the input of the second layer, and introduce a support vector discriminant item to force the sparse coding from different fingerprint points to use max- separated by margin [ 24 ]. At the same time, considering the sensitivity and instability of CSI data, the atoms in the dictionary have been proven to track the manifold structure of the training sample to overcome the influence of noise and outliers [ 25 ].…”

“… The goal of the third category is to calculate category-specific sub-dictionaries, thereby encouraging each sub-dictionary to correspond to a single category. For example, in [ 24 ], it introduces an incoherent promotion term to ensure the independence between the learned sub-dictionaries that serve a specific category. Zhou et al [ 25 ] proposed a DL algorithm that associates target categories by learning multiple dictionaries.…”

Sparse Representation Classifier Embedding Subspace Mapping and Support Vector for Facial Expression Recognition

“…We use the enhancement method to expand the data to 3 times of the original images for JAFFE and CK+ datasets. In the experiment, the SRC-SM-SV algorithm is compared with several algorithms, including SRC [12], K-SVD [18], LC-KSVD [19], FDDL [28], SVGDL [24], SDDL [21], and LCDL-SV [29]. In the SRC-SM-SV algorithm, the dimension of mapping subspace is set to be 500, and the learned dictionary has 420 atoms.…”

Locality Constrained Analysis Dictionary Learning via K-SVD Algorithm

“…Many previous works have demonstrated that local structure embedded in high-dimensional vector space is very important to characterize both the intrinsic structure and discriminative structure of images [2]. Moreover, since the data is more likely to reside on a low-dimensional sub-manifold embedded in the high-dimensional ambient space, the geometrical information of the data is important for discrimination [4,6,14]. Therefore, by preserving the locality characteristics of the training samples, the discriminative ability of the learned dictionary can be improved.…”

“…Thus, additional regularizers are incorporated in order to satisfy specific requirements in various application scenarios. For example, graph regularized sparse coding based on nonlinear manifold learning is proposed for image classing and clustering applications [3,4,11,14]. Motivated by recent progress in sparse representation, we combine the ADL model with manifold learning, called locality constrained ADL with a synthesis K-SVD solver (SK-LADL).…”