Abstract:Recent research has demonstrated the effectiveness of linear representation (i.e., sparse representation, group sparse representation and collaborative representation) for face recognition and other vision problems. However, this linear representation assumption does not consider the non-linear relationship of samples and limits the usage of different features with non-linear metrics. In this paper, we present some insights of linear and non-linear representation-based classifiers. First, we present a general formulation known as kernel collaborative representation to encompass several effective representation-based classifiers within a unified framework. Based on this framework, different algorithms can be developed by choosing proper kernel functions, regularization terms, and additional constraints. Second, within the proposed framework we develop a simple yet effective algorithm with squared ¿2-regularization and apply it to face recognition with local binary patterns as well as the Hamming kernel. We conduct numerous experiments on the extended Yale B, AR, Multi-PIE, PloyU NIR, PloyU HS, EURECOM Kinect and FERET face databases. Experimental results demonstrate that our algorithm achieves favorable performance in terms of accuracy and speed, especially for the face recognition problems with small training datasets and heavy occlusion. In addition, we attempt to combine different kernel functions by using different weights in an additive manner. The experimental results show that the proposed combination scheme provides some additional improvement in terms of accuracy. HighlightsWe present a general formulation known as kernel collaborative representation.We develop an effective algorithm with squared L2-regularization.We combine different kernels by using different weights in an additive manner.Our algorithm can achieve favorable performance in terms of accuracy and speed.

Robust Linear Embeding Based on a Kernel Function

Locally Linear Embedding Algorithm Based on Adaptive Nearest Neighbor

A modified local linear embedding algorithm based on neighbour selection.

Robust Locally Linear Embedding Algorithm for Machinery Fault Diagnosis.

Nonlinear Fault Detection Based on Locally Linear Embedding

Local Linear Transformation Embedding

Facial expression recognition based on supervised kernel local linear embedding

An Iterative Locally Linear Embedding Algorithm

Local Linear Embedding Algorithms Based on Probability Distributions

Kernel Neighborhood Preserving Projections for Face Recognition

Kernel collaborative face recognition

Variable Nearest Neighbor Locally Linear Embedding and Applications in Bearing Condition Recognition

Local Linear Embedding Algorithm Based on Geometric Distance Perturbation

Facial Expression Recognition via Locally Linear Embedding

Robust Kernel Low-Rank Representation.

Small World Neighborhood Optimized Local Linear Embedding Algorithm

Weighted Linear Embedding: Utilizing Local and Nonlocal Information Sufficiently

Kernel Relative Transformation With Applications To Enhancing Locally Linear Embedding

Improved locally linear embedding through new distance computing

Dual-weight Local Linear Embedding Algorithm Based on Adaptive Neighborhood

Clustering-Based Locally Linear Embedding