International Journal of Pattern Recognition and Artificial Intelligence

Multi-label classification has attracted significant attentions in machine learning. In multi-label classification, exploiting correlations among labels is an essential but nontrivial task. First, labels may be correlated in various degrees. Second, the scalability may suffer from the large number of labels, because the number of combinations among labels grows exponentially as the number of labels increases. In this paper, a multi-label hypernetwork (MLHN) is proposed to deal with these problems. By extending the traditional hypernetwork model, MLHN can represent arbitrary order correlations among labels. The classification model of MLHN is simple and the computational complexity of MLHN is linear with respect to the number of labels, which contribute to the good scalability of MLHN. We perform experiments on a variety of datasets. The results illustrate that the proposed MLHN achieves competitive performances against state-of-the-art multi-label classification algorithms in terms of both effectiveness and scalability with respect to the number of labels.

The goal of this work is to distinguish between humans and robots in a mixed human-robot environment. We analyze the spatio-temporal patterns of optical flow-based features along several frames. We consider the Histogram of Optical Flow (HOF) and the Motion Boundary Histogram (MBH) features, which have shown good results on people detection. The spatio-temporal patterns are composed of groups of feature components that have similar values on previous frames. The groups of features are fed into the FuzzyBoost algorithm, which at each round selects the spatio-temporal pattern (i.e. feature set) having the lowest classification error. The search for patterns is guided by grouping feature dimensions, considering three algorithms: (a) similarity of weights from dimensionality reduction matrices, (b) Boost Feature Subset Selection (BFSS) and (c) Sequential Floating Feature Selection (SFSS), which avoid the brute force approach. The similarity weights are computed by the Multiple Metric Learning for large Margin Nearest Neighbor (MMLMNN), a linear dimensionality algorithm that provides a type of Mahalanobis metric Weinberger and Saul, J. MaCh. Learn. Res.10 (2009) 207–244. The experiments show that FuzzyBoost brings good generalization properties, better than the GentleBoost, the Support Vector Machines (SVM) with linear kernels and SVM with Radial Basis Function (RBF) kernels. The classifier was implemented and tested in a real-time, multi-camera dynamic setting.

A Bayesian classifier for sparsity-promoting feature selection is developed in this paper, where a set of nonlinear mappings for the original data is performed as a pre-processing step. The linear classification model with such mappings from the original input space to a nonlinear transformation space can not only construct the nonlinear classification boundary, but also realize the feature selection for the original data. A zero-mean Gaussian prior with Gamma precision and a finite approximation of Beta process prior are used to promote sparsity in the utilization of features and nonlinear mappings in our model, respectively. We derive the Variational Bayesian (VB) inference algorithm for the proposed linear classifier. Experimental results based on the synthetic data set, measured radar data set, high-dimensional gene expression data set, and several benchmark data sets demonstrate the aggressive and robust feature selection capability and comparable classification accuracy of our method comparing with some other existing classifiers.

Multidimensional scaling techniques are unsupervised Dimension Reduction (DR) techniques which use multidimensional data pairwise similarities to represent data into a plane enabling their visual exploratory analysis. Considering labeled data, the DR techniques face two objectives with potentially different priorities: one is to account for the data points' similarities, the other for the data classes' structures. Unsupervised DR techniques attempt to preserve original data similarities, but they do not consider their class label hence they can map originally separated classes as overlapping ones. Conversely, the state-of-the-art so-called supervised DR techniques naturally handle labeled data, but they do so in a predictive modeling framework where they attempt to separate the classes in order to improve a classification accuracy measure in the low-dimensional space, hence they can map as separated even originally overlapping classes. We propose ClassiMap, a DR technique which optimizes a new objective function enabling Exploratory Data Analysis (EDA) of labeled data. Mapping distortions known as tears and false neighborhoods cannot be avoided in general due to the reduction of the data dimension. ClassiMap intends primarily to preserve data similarities but tends to distribute preferentially unavoidable tears among the different-label data and unavoidable false neighbors among the same-label data. Standard quality measures to evaluate the quality of unsupervised mappings cannot tell about the preservation of within-class or between-class structures, while classification accuracy used to evaluate supervised mappings is only relevant to the framework of predictive modeling. We propose two measures better suited to the evaluation of DR of labeled data in an EDA framework. We use these two label-aware indices and four other standard unsupervised indices to compare ClassiMap to other state-of-the-art supervised and unsupervised DR techniques on synthetic and real datasets. ClassiMap appears to provide a better tradeoff between pairwise similarities and class structure preservation according to these new measures.

The presence of noise, loss of information or feature nonstationarity in data is the limiting factor for many machine learning decision systems. Previous research has shown that relevant feature selection may be helpful to alleviate the impact of these possible perturbations. This paper presents a dynamical feature subspaces selection method based on ensembles of one-class Support Vector Machine (SVM), with the objective to optimize the performance of a decision system in such a nonstationary environment. Our method is predicated on the assumption that only the performance of the classifiers using perturbed features is degraded. We propose a mechanism for constructing an ensemble of classifiers based on a large number of feature subspaces generated from the initial full-dimensional space. In the phase of classification, the ensemble system is capable to select adaptively feature subspaces which are supposed to be immune to the nonstationary disturbance and to make the final decision by combining the individual decisions of classifiers built in these subspaces. One characteristic of this method is that we use the one-class SVM ensemble to accomplish simultaneously the tasks of feature subspace selection and classification. The effectiveness of the proposed method has been demonstrated through the experiments conducted in the context of textured image classification.

In recent years, the theory of sparse representation (SR) has been widely exploited in sparse subspace learning (SSL). Among all these methods, SR is a parameter-free global algorithm in nature which is mostly utilized to construct the correlations between samples to avoid some negative effects incurred by k-nearest neighbor (KNN) or some other methods. However, these SSL algorithms always lack obvious discrimination because of the ignorance of samples distribution. Meanwhile, some incorrect correlations are taken into consideration owing to the global feature of SR. To solve these two problems, a new SSL algorithm called locality structured sparsity preserving embedding (LSPE) is proposed in this paper. We add the local structured information to SR and construct correlations between samples. However, LSPE is an unsupervised method which wastes all label information. Therefore, LSPE is extended to semi-supervised LSPE (SLSPE) in this paper. SLSPE not only makes good use of the label information but also enhances the discriminative power of LSPE. Extensive experiments have been performed on three image datasets (CMU, COIL20, ORL) and two UCI datasets (Glass, Segment) to prove the efficiency of the LSPE and SLSPE.

Keyword-based image retrieval is more comfortable for users than content-based image retrieval. Because of the lack of semantic description of images, image annotation is often used a priori by learning the association between the semantic concepts (keywords) and the images (or image regions). This association issue is particularly difficult but interesting because it can be used for annotating images but also for multimodal image retrieval. However, most of the association models are unidirectional, from image to keywords. In addition to that, existing models rely on a fixed image database and prior knowledge. In this paper, we propose an original association model, which provides image-keyword bidirectional transformation. Based on the state-of-the-art Bag of Words model dealing with image representation, including a strategy of interactive incremental learning, our model works well with a zero-or-weak-knowledge image database and evolving from it. Some objective quantitative and qualitative evaluations of the model are proposed, in order to highlight the relevance of the method.

In this paper, we propose a new approach for anomaly detection in video surveillance. This approach is based on a nonparametric Bayesian regression model built upon Gaussian process priors. It establishes a set of basic vectors describing motion patterns from low-level features via online clustering, and then constructs a Gaussian process regression model to approximate the distribution of motion patterns in kernel space. We analyze different anomaly measure criterions derived from Gaussian process regression model and compare their performances. To reduce false detections caused by crowd occlusion, we utilize supplement information from previous frames to assist in anomaly detection for current frame. In addition, we address the problem of hyperparameter tuning and discuss the method of efficient calculation to reduce computation overhead. The approach is verified on published anomaly detection datasets and compared with other existing methods. The experiment results demonstrate that it can detect various anomalies efficiently and accurately.

Decision making is one of the smouldering problems in day to day works. Human emotions play crucial role in decision-making systems. While person is in high emotion he cannot make proper decision. Robust local binary pattern (RLBP) operator is more powerful to recognize the emotions and extends the features of local binary pattern (LBP). However, there are some precincts like discriminating bright faces against dark features and vice versa and intra-class variances increase. The RLBP solves this problem by finding minimum of LBP codes and their complements. However, it miss the mark for different local structures a similar feature is obtained, weak contrast local patterns and similar strong contrast local patterns. Hence, the discriminative robust local binary pattern (DRLBP) method is proposed to retain the contrast information of image patterns next to considering both edge and texture information. Nevertheless, LBP family methods are highly sensitive to noise. To trounce these drawbacks this paper extends fuzzy rule-based DRLBP which is more robust to noise, low contrasted, uneven lighting conditions, variations in expressions and rotation variant images.

This paper proposes hybrid approaches based on both feature level and score level fusion strategies to provide a robust recognition system against the distortions of individual modalities. In order to compare the proposed schemes, a virtual multimodal database is formed from FERET face and PolyU palmprint databases. The proposed hybrid systems concatenate features extracted by local and global feature extraction methods such as Local Binary Patterns, Log Gabor, Principal Component Analysis and Linear Discriminant Analysis. Match score level fusion is performed in order to show the effectiveness and accuracy of the proposed schemes. The experimental results based on these databases reported a significant improvement of the proposed schemes compared with unimodal systems and other multimodal face–palmprint fusion methods.

In attempting to overcome the limitation of current methods to solve complicated constrained optimization problems, this paper proposes an adaptive hybrid particle swarm optimization multi-objective optimization (AHPSOMO) algorithm. In the early stage, this algorithm initializes the individuals in a population in an even manner using good point set (GPS) theory so that the diversity of the population can be guaranteed. In the process of local search, differential evolution (DE) algorithm is introduced for updating local optimal individuals. Particle swarm optimization method is further adopted to conduct global search as per the multi-objective approach. The results of simulation tests on 24 classic test functions and three engineering constrained optimization problems show that compared with other algorithms, our proposed algorithm is effective and feasible, which can offer highly accurate solutions with good robustness.

A novel strength Pareto gravitational search algorithm (SPGSA) is proposed to solve multi-objective optimization problems. This SPGSA algorithm utilizes the strength Pareto concept to assign the fitness values for agents and uses a fine-grained elitism selection mechanism to keep the population diversity. Furthermore, the recombination operators are modeled in this approach to decrease the possibility of trapping in local optima. Experiments are conducted on a series of benchmark problems that are characterized by difficulties in local optimality, nonuniformity, and nonconvexity. The results show that the proposed SPGSA algorithm performs better in comparison with other related works. On the other hand, the effectiveness of two subtle means added to the GSA are verified, i.e. the fine-grained elitism selection and the use of SBX and PMO operators. Simulation results show that these measures not only improve the convergence ability of original GSA, but also preserve the population diversity adequately, which enables the SPGSA algorithm to have an excellent ability that keeps a desirable balance between the exploitation and exploration so as to accelerate the convergence speed to the true Pareto-optimal front.