Xianglei Xing

Complete canonical correlation analysis with application to multi-view gait recognition

Canonical correlation analysis (CCA) is a well-known multivariate analysis method for quantifying the correlations between two sets of multidimensional variables. However, for multi-view gait recognition, it is difficult to directly apply CCA to deal with two sets of high-dimensional vectors because of computational complexity. Moreover, in such situation, the eigenmatrix of CCA is usually singular which makes the direct implementation of the CCA algorithm almost impossible. In practice, PCA or singular value decomposition is employed as a preprocessing step to solve these problems. Nevertheless, this strategy may discard dimensions that contain important discriminative information and correlation information. To overcome the shortcomings of CCA when dealing with two sets of high-dimensional vectors, we develop a novel method, named complete canonical correlation analysis (C3A). In our method, we first reformulate the traditional CCA so that we can avoid the computing of the inverse of a high-dimensional matrix. With the help of this reformulation, C3A further transforms the singular generalized eigensystem computation of CCA into two stable eigenvalue decomposition problems. Moreover, a feasible and effective method is proposed to alleviate the computational burden of high dimensional matrix for typical gait image data. Experimental results on two benchmark gait databases, CASIA gait database and the challenge USF gait database, demonstrate the effectiveness of the proposed method. HighlightsWe overcome the shortcomings of CCA when dealing with high-dimensional matrix.The singularity of generalized eigenvalue problem in CCA is overcome naturally.The important discriminative information is preserved completely in our algorithm.Our scheme learns stable and complete solutions.The multi-view gait recognition is achieved based on our method.

Class Energy Image Analysis for Video Sensor-Based Gait Recognition: A Review

Gait is a unique perceptible biometric feature at larger distances, and the gait representation approach plays a key role in a video sensor-based gait recognition system. Class Energy Image is one of the most important gait representation methods based on appearance, which has received lots of attentions. In this paper, we reviewed the expressions and meanings of various Class Energy Image approaches, and analyzed the information in the Class Energy Images. Furthermore, the effectiveness and robustness of these approaches were compared on the benchmark gait databases. We outlined the research challenges and provided promising future directions for the field. To the best of our knowledge, this is the first review that focuses on Class Energy Image. It can provide a useful reference in the literature of video sensor-based gait representation approach.

Fusion of Gait and Facial Features using Coupled Projections for People Identification at a Distance

A novel feature-level fusion scheme for people identification at a distance has been developed by coupling gait feature with facial feature. The proposed coupled projections based method first maps the heterogeneous features from gait and face into a unified subspace to minimize the distance between the two features extracted from the same individual. The fusion features are obtained by computing the mean of the two projecting features from the same person in the coupled subspace. Experimental results demonstrate that the proposed feature-level fusion scheme outperforms the match score-level and two other feature-level fusion schemes in the application of access control at a distance.

Fusion of Local Manifold Learning Methods

Different local manifold learning methods are developed based on different geometric intuitions and each method only learns partial information of the true geometric structure of the underlying manifold. In this letter, we introduce a novel method to fuse the geometric information learned from local manifold learning algorithms to discover the underlying manifold structure more faithfully. We first use local tangent coordinates to compute the local objects from different local algorithms, then utilize the selection matrix to connect the local objects with a global functional and finally develop an alternating optimization-based algorithm to discover the low-dimensional embedding. Experiments on synthetic as well as real datasets demonstrate the effectiveness of our proposed method.

Multi-step-ahead host load prediction using autoencoder and echo state networks in cloud computing

Cloud computing is a model for enabling ubiquitous, convenient, on-demand network access to a shared pool of configurable computing resources that can be rapidly provisioned and released with minimal management effort or service provider interaction. There are many proposals for resource management approaches for cloud infrastructures, but effective resource management is still a major challenge for the leading cloud infrastructure operators (e.g., Amazon, Microsoft, Google), because the details of the underlying workloads and the real-world operational demands are too complex. Among those proposals, accurate host load prediction is one of the most effective measures to address this challenge. In this paper, we proposed a new method for host load prediction, which uses an autoencoder as the pre-recurrent feature layer of the echo state networks. The aim of our proposed method is to predict the host load in the future interval based on Google cluster usage dataset. Experiments performed on Google load traces show that our proposed method achieves higher accuracy than the state-of-the-art methods.

Robust Hessian Locally Linear Embedding Techniques for High-Dimensional Data

Recently manifold learning has received extensive interest in the community of pattern recognition. Despite their appealing properties, most manifold learning algorithms are not robust in practical applications. In this paper, we address this problem in the context of the Hessian locally linear embedding (HLLE) algorithm and propose a more robust method, called RHLLE, which aims to be robust against both outliers and noise in the data. Specifically, we first propose a fast outlier detection method for high-dimensional datasets. Then, we employ a local smoothing method to reduce noise. Furthermore, we reformulate the original HLLE algorithm by using the truncation function from differentiable manifolds. In the reformulated framework, we explicitly introduce a weighted global functional to further reduce the undesirable effect of outliers and noise on the embedding result. Experiments on synthetic as well as real datasets demonstrate the effectiveness of our proposed algorithm.

Erratum: Couple Metric Learning Based on Separable Criteria with Its Application in Cross-View Gait Recognition

Gait is an important biometric feature to identify a person at a distance. However, the performance of the traditional gait recognition methods may degenerate when the viewing angle is changed. This is because the viewing angle of the probe data may not be the same as the viewing angle under which the gait signature database is generated. In this paper, we introduce the separable criteria into the couple metric learning (CML) method, and apply this novel method to normalize gait features from various viewing angles into a couple feature spaces. Then, the gait similarity measurement is conducted in this common feature space. We incorporate the label information into the separable criteria to improve the performance of the traditional CML method. Experiments are performed on the benchmark gait database. The results demonstrate the efficiency of our method.

Decision-Level Fusion Method Based on Deep Learning

We present a highly accurate and very efficient approach for personality traits prediction based on video. Unlike the traditional method, we proposed a decision-level information fusion method based on deep learning. We have separated the video modal into two parts, visual modal and audio model. The two models were processed by improved VGG-16 and LSTM network, respectively, and combined with an Extreme Learning Machine (ELM) to architecture decision-level information fusion. Experiments on challenging Youtube-8M dataset show that our proposed approach significantly outperforms traditional decision-level fusion method in terms of both efficiency and accuracy.

Contrast Research on Full Finger Area Extraction Method of Touchless Fingerprint Images Under Different Illuminants.

Touchless fingerprint recognition with high acceptance, high security, hygiene advantages, is currently a hot research field of biometrics. The background areas of touchless fingerprints are more complex and bigger than those of the contact. So the general methods for contact fingerprint images are difficult to achieve a good effect when extracting the full finger area. The purpose of this research is to compare the performance of finger area extraction based on different color model and illuminants, and then lays the foundation for touchless fingerprint identification. The fingerprint images are respectively collected under blue, green and red illuminants. And then, the Otsu based on YCbCr model, HSV model, and YIQ model is adopted to extract the finger area. Experimental results show that the Otsu based on the Cb component of YCbCr model and S component of HSV model can achieve excellent extraction results under blue illuminant.

Feature Level Information Fusion Based Deep Learning

Encouraged by recent methods disable to achieve good tradeoff between accuracy and convergence. To close the gap, we propose to combine multi-feature based deep learning. We enable our analysis by facial recognition and comparison. We increase proportion of face and feature in an image. Firstly, we crop face,eyes, nose and mouth regions. Second, we extract features and combine them. It can be shown that it is efficient and it has capable of convergence quickly in facial recognition. Our method achieves the best performance on LWF by 97.98%. We make facial comparison by improved Siamese network. In the network, we add Spatial Transformer Networks. With improved Siamese network, it can be efficiently optimized with different perspectives and thus guarantee good robustness. Extensive experiments demonstrate that accuracy and stability improve significantly than tradition Siamese network. Furthermore, our method has good generalization. Without training again when you want to compare two images. These algorithms implanted to C# platform, we make interface of facial recognition and comparison.