Minnan Luo

An Adaptive Semisupervised Feature Analysis for Video Semantic Recognition

Video semantic recognition usually suffers from the curse of dimensionality and the absence of enough high-quality labeled instances, thus semisupervised feature selection gains increasing attentions for its efficiency and comprehensibility. Most of the previous methods assume that videos with close distance (neighbors) have similar labels and characterize the intrinsic local structure through a predetermined graph of both labeled and unlabeled data. However, besides the parameter tuning problem underlying the construction of the graph, the affinity measurement in the original feature space usually suffers from the curse of dimensionality. Additionally, the predetermined graph separates itself from the procedure of feature selection, which might lead to downgraded performance for video semantic recognition. In this paper, we exploit a novel semisupervised feature selection method from a new perspective. The primary assumption underlying our model is that the instances with similar labels should have a larger probability of being neighbors. Instead of using a predetermined similarity graph, we incorporate the exploration of the local structure into the procedure of joint feature selection so as to learn the optimal graph simultaneously. Moreover, an adaptive loss function is exploited to measure the label fitness, which significantly enhances model’s robustness to videos with a small or substantial loss. We propose an efficient alternating optimization algorithm to solve the proposed challenging problem, together with analyses on its convergence and computational complexity in theory. Finally, extensive experimental results on benchmark datasets illustrate the effectiveness and superiority of the proposed approach on video semantic recognition related tasks.

Adaptive Unsupervised Feature Selection With Structure Regularization

Feature selection is one of the most important dimension reduction techniques for its efficiency and interpretation. Since practical data in large scale are usually collected without labels, and labeling these data are dramatically expensive and time-consuming, unsupervised feature selection has become a ubiquitous and challenging problem. Without label information, the fundamental problem of unsupervised feature selection lies in how to characterize the geometry structure of original feature space and produce a faithful feature subset, which preserves the intrinsic structure accurately. In this paper, we characterize the intrinsic local structure by an adaptive reconstruction graph and simultaneously consider its multiconnected-components (multicluster) structure by imposing a rank constraint on the corresponding Laplacian matrix. To achieve a desirable feature subset, we learn the optimal reconstruction graph and selective matrix simultaneously, instead of using a predetermined graph. We exploit an efficient alternative optimization algorithm to solve the proposed challenging problem, together with the theoretical analyses on its convergence and computational complexity. Finally, extensive experiments on clustering task are conducted over several benchmark data sets to verify the effectiveness and superiority of the proposed unsupervised feature selection algorithm.

Simple to complex cross-modal learning to rank

Abstract The heterogeneity-gap between different modalities brings a significant challenge to multimedia information retrieval. Some studies formalize the cross-modal retrieval tasks as a ranking problem and learn a shared multi-modal embedding space to measure the cross-modality similarity. However, previous methods often establish the shared embedding space based on linear mapping functions which might not be sophisticated enough to reveal more complicated inter-modal correspondences. Additionally, current studies assume that the rankings are of equal importance, and thus all rankings are used simultaneously, or a small number of rankings are selected randomly to train the embedding space at each iteration. Such strategies, however, always suffer from outliers as well as reduced generalization capability due to their lack of insightful understanding of procedure of human cognition. In this paper, we involve the self-paced learning theory with diversity into the cross-modal learning to rank and learn an optimal multi-modal embedding space based on non-linear mapping functions. This strategy enhances the model’s robustness to outliers and achieves better generalization via training the model gradually from easy rankings by diverse queries to more complex ones. An efficient alternative algorithm is exploited to solve the proposed challenging problem with fast convergence in practice. Extensive experimental results on several benchmark datasets indicate that the proposed method achieves significant improvements over the state-of-the-arts in this literature.

Avoiding optimal mean ź2,1-norm maximization-based robust pca for reconstruction

Robust principal component analysis (PCA) is one of the most important dimension-reduction techniques for handling high-dimensional data with outliers. However, most of the existing robust PCA presupposes that the mean of the data is zero and incorrectly utilizes the average of data as the optimal mean of robust PCA. In fact, this assumption holds only for the squared -norm-based traditional PCA. In this letter, we equivalently reformulate the objective of conventional PCA and learn the optimal projection directions by maximizing the sum of projected difference between each pair of instances based on -norm. The proposed method is robust to outliers and also invariant to rotation. More important, the reformulated objective not only automatically avoids the calculation of optimal mean and makes the assumption of centered data unnecessary, but also theoretically connects to the minimization of reconstruction error. To solve the proposed nonsmooth problem, we exploit an efficient optimization algorithm to soften the contributions from outliers by reweighting each data point iteratively. We theoretically analyze the convergence and computational complexity of the proposed algorithm. Extensive experimental results on several benchmark data sets illustrate the effectiveness and superiority of the proposed method.

Robust dictionary learning with graph regularization for unsupervised person re-identification

Most existing approaches for person re-identification are designed in a supervised way, undergoing a prohibitively high labeling cost and poor scalability. Besides establishing effective similarity distance metrics, these supervised methods usually focus on constructing discriminative and robust features, which is extremely difficult due to the significant viewpoint variations. To overcome these challenges, we propose a novel unsupervised method, termed as Robust Dictionary Learning with Graph Regularization (RDLGR), which can guarantee view-invariance through learning a dictionary shared by all the camera views. To avoid the significant degradation of performance caused by outliers, we employ a capped l2,1-norm based loss to make our model more robust, addressing the problem that traditional quadratic loss is known to be easily dominated by outliers. Considering the lack of labeled cross-view discriminative information in our unsupervised method, we further introduce a cross-view graph Laplacian regularization term into the framework of dictionary learning. As a result, the geographical structure of original data space can be preserved in the learned latent subspace as discriminative information, making it possible to further boost the matching accuracy. Extensive experimental results over four widely used benchmark datasets demonstrate the superiority of the proposed model over the state-of-the-art methods.

Sparse Relational Topical Coding on multi-modal data

Abstract Multi-modal data modeling lately has been an active research area in pattern recognition community. Existing studies mainly focus on modeling the content of multi-modal documents, whilst the links amongst documents are commonly ignored. However, link information has shown being of key importance in many applications, such as document navigation, classification, and clustering. In this paper, we present a non-probabilistic formulation of Relational Topic Model (RTM), i.e., Sparse Relational Multi-Modal Topical Coding (SRMMTC), to model both multi-modal documents and the corresponding link information. SRMMTC has the following three appealing properties: i) It can effectively produce sparse latent representations via directly imposing sparsity-inducing regularizers. ii) It handles the imbalance issues on multi-modal data collections by introducing regularization parameters for positive and negative links, respectively; iii) It can be solved by an efficient coordinate descent algorithm. We also explore a generalized version of SRMMTC to find pairwise interactions amongst topics. Our methods are also capable of performing link prediction for documents, as well as the prediction of annotation words for attendant images in documents. Empirical studies on a set of benchmark datasets show that our proposed models significantly outperform many state-of-the-art methods.

Distributed Extreme Learning Machine with Alternating Direction Method of Multiplier

Extreme learning machine, as a generalized single-hidden-layer feedforward networks has achieved much attention for its extremely fast learning speed and good generalization performance. However, big data often makes a challenge in large scale learning of ELM due to the limitation of memory of single machine as well as the distributed manner of large scale data storage and collection in many applications. For the purpose of relieving the limitation of memory with big data, in this paper, we exploit a novel distributed extreme learning machine to implement the extreme learning machine algorithm in parallel for large-scale data set. A corresponding distributed algorithm is also developed on the basis of alternating direction method of multipliers which shows effectiveness in distributed convex optimization. Finally, some numerical experiments on well-known benchmark data sets are carried out to illustrate the effectiveness of the proposed DELM method and provide an analysis on the performance of speedup, scaleup and sizeup.

Sparse Multi-Modal Topical Coding for Image Annotation

Image annotation plays a significant role in large scale image understanding, indexing and retrieval. The Probability Topic Models (PTMs) attempt to address this issue by learning latent representations of input samples, and have been shown to be effective by existing studies. Though useful, PTM has some limitations in interpreting the latent representations of images and texts, which if addressed would broaden its applicability. In this paper, we introduce sparsity to PTM to improve the interpretability of the inferred latent representations. Extending the Sparse Topical Coding that originally designed for unimodal documents learning, we propose a non-probabilistic formulation of PTM for automatic image annotation, namely Sparse Multi-Modal Topical Coding. Beyond controlling the sparsity, our model can capture more compact correlations between words and image regions. Empirical results on some benchmark datasets show that our model achieves better performance on automatic image annotation and text-based image retrieval over the baseline models.

ANOMALOUS: A Joint Modeling Approach for Anomaly Detection on Attributed Networks

The key point of anomaly detection on attributed networks lies in the seamless integration of network structure information and attribute information. A vast majority of existing works are mainly based on the Homophily assumption that implies the nodal attribute similarity of connected nodes. Nonetheless, this assumption is untenable in practice as the existence of noisy and structurally irrelevant attributes may adversely affect the anomaly detection performance. Despite the fact that recent attempts perform subspace selection to address this issue, these algorithms treat subspace selection and anomaly detection as two separate steps which often leads to suboptimal solutions. In this paper, we investigate how to fuse attribute and network structure information more synergistically to avoid the adverse effects brought by noisy and structurally irrelevant attributes. Methodologically, we propose a novel joint framework to conduct attribute selection and anomaly detection as a whole based on CUR decomposition and residual analysis. By filtering out noisy and irrelevant node attributes, we perform anomaly detection with the remaining representative attributes. Experimental results on both synthetic and real-world datasets corroborate the effectiveness of the proposed framework.

Deep Semisupervised Zero-Shot Learning with Maximum Mean Discrepancy

Due to the difficulty of collecting labeled images for hundreds of thousands of visual categories, zero-shot learning, where unseen categories do not have any labeled images in training stage, has attracted more attention. In the past, many studies focused on transferring knowledge from seen to unseen categories by projecting all category labels into a semantic space. However, the label embeddings could not adequately express the semantics of categories. Furthermore, the common semantics of seen and unseen instances cannot be captured accurately because the distribution of these instances may be quite different. For these issues, we propose a novel deep semisupervised method by jointly considering the heterogeneity gap between different modalities and the correlation among unimodal instances. This method replaces the original labels with the corresponding textual descriptions to better capture the category semantics. This method also overcomes the problem of distribution difference by minimizing the maximum mean discrepancy between seen and unseen instance distributions. Extensive experimental results on two benchmark data sets, CU200-Birds and Oxford Flowers-102, indicate that our method achieves significant improvements over previous methods.