Semantic preserving distance metric learning and applications

Harnessing Side Information for Classification Under Label Noise

Practical data sets often contain the label noise caused by various human factors or measurement errors, which means that a fraction of training examples might be mistakenly labeled. Such noisy labels will mislead the classifier training and severely decrease the classification performance. Existing approaches to handle this problem are usually developed through various surrogate loss functions under the framework of empirical risk minimization. However, they are only suitable for binary classification and also require strong prior knowledge. Therefore, this article treats the example features as side information and formulates the noisy label removal problem as a matrix recovery problem. We denote our proposed method as "label noise handling via side information" (LNSI). Specifically, the observed label matrix is decomposed as the sum of two parts, in which the first part reveals the true labels and can be obtained by conducting a low-rank mapping on the side information; and the second part captures the incorrect labels and is modeled by a row-sparse matrix. The merits of such formulation lie in three aspects: 1) the strong recovery ability of this strategy has been sufficiently demonstrated by intensive theoretical works on side information; 2) multi-class situations can be directly handled with the aid of learned projection matrix; and 3) only very weak assumptions are required for model design, making LNSI applicable to a wide range of practical problems. Moreover, we theoretically derive the generalization bound of LNSI and show that the expected classification error of LNSI is upper bounded. The experimental results on a variety of data sets including UCI benchmark data sets and practical data sets confirm the superiority of LNSI to state-of-the-art approaches on label noise handling.

Content-based retrieval for lung nodule diagnosis using learned distance metric

Similarity metric of the lung nodules can be useful in differentiating between benign and malignant lung nodule lesions on computed tomography (CT). Unlike previous computerized schemes, which focus on the features extracting, we concentrate on similarity metric of the lung nodules. In this study, we first assemble a lung nodule dataset which is from LIDC-IDRI lung CT images. This dataset includes 746 lung nodules in which 375 domain radiologists identified malignant nodules and 371 domain radiologists-identified benign nodules. Each nodule is represented by a vector of 26 texture features. We then propose a content-based image retrieval (CBIR) scheme to classify between benign and malignant lung nodules with a learned Mahalanobis distance metric. The Mahalanobis distance metric as a similarity metric can preserve semantic relevance and visual similarity of lung nodules. The CBIR approach uses this Mahalanobis distance to search for most similar reference nodules for each queried nodule. The majority of votes are then computed to predict the likelihood of the queried nodule depicting a malignant lesion. For the classification accuracy, the area under the ROC curve (AUC) can achieve as 0.942±0.008. The recall and precision of benign nodules are 0.860 and 0.889, respectively. The recall and precision of malignant nodules are 0.893 and 0.866, respectively.

Tensor Rank Preserving Discriminant Analysis for Facial Recognition

Facial recognition, one of the basic topics in computer vision and pattern recognition, has received substantial attention in recent years. However, for those traditional facial recognition algorithms, the facial images are reshaped to a long vector, thereby losing part of the original spatial constraints of each pixel. In this paper, a new tensor-based feature extraction algorithm termed tensor rank preserving discriminant analysis (TRPDA) for facial image recognition is proposed; the proposed method involves two stages: in the first stage, the low-dimensional tensor subspace of the original input tensor samples was obtained; in the second stage, discriminative locality alignment was utilized to obtain the ultimate vector feature representation for subsequent facial recognition. On the one hand, the proposed TRPDA algorithm fully utilizes the natural structure of the input samples, and it applies an optimization criterion that can directly handle the tensor spectral analysis problem, thereby decreasing the computation cost compared those traditional tensor-based feature selection algorithms. On the other hand, the proposed TRPDA algorithm extracts feature by finding a tensor subspace that preserves most of the rank order information of the intra-class input samples. Experiments on the three facial databases are performed here to determine the effectiveness of the proposed TRPDA algorithm.

Large Sparse Cone Non-negative Matrix Factorization for Image Annotation

Image annotation assigns relevant tags to query images based on their semantic contents. Since Non-negative Matrix Factorization (NMF) has the strong ability to learn parts-based representations, recently, a number of algorithms based on NMF have been proposed for image annotation and have achieved good performance. However, most of the efforts have focused on the representations of images and annotations. The properties of the semantic parts have not been well studied. In this article, we revisit the sparseness-constrained NMF (sNMF) proposed by Hoyer [2004]. By endowing the sparseness constraint with a geometric interpretation and sNMF with theoretical analyses of the generalization ability, we show that NMF with such a sparseness constraint has three advantages for image annotation tasks: (i) The sparseness constraint is more ℓ 0 -norm oriented than the ℓ 1 -norm-based sparseness, which significantly enhances the ability of NMF to robustly learn semantic parts. (ii) The sparseness constraint has a large cone interpretation and thus allows the reconstruction error of NMF to be smaller, which means that the learned semantic parts are more powerful to represent images for tagging. (iii) The learned semantic parts are less correlated, which increases the discriminative ability for annotating images. Moreover, we present a new efficient large sparse cone NMF (LsCNMF) algorithm to optimize the sNMF problem by employing the Nesterov’s optimal gradient method. We conducted experiments on the PASCAL VOC07 dataset and demonstrated the effectiveness of LsCNMF for image annotation.

Similarity measurement of lung masses for medical image retrieval using kernel based semisupervised distance metric

PURPOSE

To develop a new algorithm to measure the similarity between the query lung mass and reference lung mass data set for content-based medical image retrieval (CBMIR).

METHODS

A lung mass data set including 746 mass regions of interest (ROIs) was assembled. Among them, 375 ROIs depicted malignant lesions and 371 depicted benign lesions. Each mass ROI is represented by a vector of 26 texture features. A kernel function was employed to map the original data in input space to a feature space. In this space, a semisupervised distance metric was learned, which used differential scatter discriminant criterion to represent the semantic relevance, and the regularization term to represent the visual similarity. The learned distance metric can measure the similarity of the query mass and reference mass data set. The clustering accuracy is used to configure the parameters. The retrieval accuracy and classification accuracy are used as the performance assessment index.

RESULTS

After configuring the parameters, a mean clustering accuracy of 0.87 can be achieved. For retrieval accuracy, our algorithm achieves better performance than other state-of-the-art retrieval algorithms when applying a leave-one-out validation method to the testing data set. For classification accuracy, the area under the ROC curve of our algorithm can be achieved as 0.941 ± 0.006. The running times of 346 query images with the proposed algorithm are 5.399 and 6.0 s, respectively.

CONCLUSIONS

The study results demonstrated the proposed algorithm outperforms the compared algorithms, when taking the semantic relevant and visual similarity into account in kernel space. The algorithm can be used in a CBMIR system for a query mass to retrieve similarity masses, which can help doctors make better decisions.

Latent Max-Margin Multitask Learning With Skelets for 3-D Action Recognition

Recent emergence of low-cost and easy-operating depth cameras has reinvigorated the research in skeleton-based human action recognition. However, most existing approaches overlook the intrinsic interdependencies between skeleton joints and action classes, thus suffering from unsatisfactory recognition performance. In this paper, a novel latent max-margin multitask learning model is proposed for 3-D action recognition. Specifically, we exploit skelets as the mid-level granularity of joints to describe actions. We then apply the learning model to capture the correlations between the latent skelets and action classes each of which accounts for a task. By leveraging structured sparsity inducing regularization, the common information belonging to the same class can be discovered from the latent skelets, while the private information across different classes can also be preserved. The proposed model is evaluated on three challenging action data sets captured by depth cameras. Experimental results show that our model consistently achieves superior performance over recent state-of-the-art approaches.