Dictionary Space Research Articles

Wasserstein Discriminant Dictionary Learning for Graph Representation.

Mining discriminative graph topological information plays an important role in promoting graph representation ability. However, it suffers from two main issues: (1) the difficulty/complexity of computing global inter-class/intra-class scatters, commonly related to mean and covariance of graph samples, for discriminant learning; (2) the huge complexity and variety of graph topological structure that is rather challenging to robustly characterize. In this paper, we propose the Wasserstein Discriminant Dictionary Learning (WDDL) framework to achieve discriminant learning on graphs with robust graph topology modeling, and hence facilitate graph-based pattern analysis tasks. Considering the difficulty of calculating global inter-class/intra-class scatters, a reference set of graphs (aka graph dictionary) is first constructed by generating representative graph samples (aka graph keys) with expressive topological structure. Then, a Wasserstein Graph Representation (WGR) process is proposed to project input graphs into a succinct dictionary space through the graph dictionary lookup. To further achieve discriminant graph learning, a Wasserstein discriminant loss (WD-loss) is defined on the graph dictionary, in which the graph keys are optimizable, to make the intra-class keys more compact and inter-class keys more dispersed. Hence, the calculation of global Wasserstein metric (W-metric) centers can be bypassed. For sophisticated topology mining in the WGR process, a joint-Wasserstein graph embedding module is constructed to model both between-node and between-edge relationships across inputs and graph keys by encapsulating both the Wasserstein metric (between cross-graph nodes) and proposed novel Kron-Gromov-Wasserstein (KGW) metric (between cross-graph adjacencies). Specifically, the KGW-metric comprehensively characterizes the cross-graph connection patterns with the Kronecker operation, then adaptively captures those salient patterns through connection pooling. To evaluate the proposed framework, we study two graph-based pattern analysis problems, i.e. graph classification and cross-modal retrieval, with the graph dictionary flexibly adjusted to cater to these two tasks. Extensive experiments are conducted to comprehensively compare with existing advanced methods, as well as dissect the critical component of our proposed architecture. The experimental results validate the effectiveness of the WDDL framework.

An improved orthogonal matching pursuit method for denoising high-frequency ultrasonic detection signals of flip chips

Noise suppression of an echo signal plays an important role in high-frequency ultrasonic testing of flip chips. This paper proposes an orthogonal matching pursuit (OMP) method optimized by an improved artificial bee colony (ABC) algorithm for denoising high-frequency ultrasonic testing signals of flip chips. We add adaptive learning factors to change the way the ABC randomly selects the search direction, which speeds up the convergence speed of the algorithm. The improved ABC, named adaptive artificial bee colony (AABC), replaces the greedy search process of OMP. Instead of searching for the atom that best matches the echo signal, the improved OMP algorithm searches for the optimal parameter and replaces the atom with a set of parameters with practical physical significance. The introduction of the AABC changes the search space of OMP from discrete dictionary space to continuous parameter space, which minimizes the error between the atom and echo signal and leads to an accurate approximation. Additionally, the AABC reduces the search times of OMP and improves the convergence speed. Then, the wavelet transform thresholding technology is combined with the attenuation characteristic of high-frequency ultrasound to eliminate the influence of error decomposition caused by noise and inaccurate sparsity setting. We present the noise suppression experimental results of high-frequency ultrasonic simulation and real testing signals of flip chips, including Gaussian white noise and correlated noise, and compare the proposed method with other sparse representation methods, including matching pursuit (MP) and OMP. The results demonstrate the superior performance of the proposed method.

High-Quality MR Fingerprinting Reconstruction Using Structured Low-Rank Matrix Completion and Subspace Projection.

Due to the capability of fast multiparametric quantitative imaging, magnetic resonance fingerprinting (MRF) is becoming a promising quantitative magnetic resonance imaging approach. However, the artifacts caused by the highly undersampled data acquisition lead to inaccurate estimation of the tissue parameter maps. Based on the assumption that the 3-D MRF data can be modeled as a piecewise smooth signal, with the discontinuities localized to the zero sets of a bandlimited function, we exploit the low-rank property of the structured Toeplitz matrix constructed from the Fourier measurements. In addition, we adopt the subspace projection scheme to improve the accuracy of parameter estimation. In order to efficiently solve the regularized problem, we propose an iterative two-stage algorithm, which alternately updates the k -space data and projects the space-time matrix into the dictionary space. Numerical experiments demonstrate that the proposed algorithm shows significant improvement in MRF time-series images reconstruction and can provide more accurate parameter maps over the state-of-the-art algorithms.

Deep Wasserstein Graph Discriminant Learning for Graph Classification

Graph topological structures are crucial to distinguish different-class graphs. In this work, we propose a deep Wasserstein graph discriminant learning (WGDL) framework to learn discriminative embeddings of graphs in Wasserstein-metric (W-metric) matching space. In order to bypass the calculation of W-metric class centers in discriminant analysis, as well as better support batch process learning, we introduce a reference set of graphs (aka graph dictionary) to express those representative graph samples (aka dictionary keys). On the bridge of graph dictionary, every input graph can be projected into the latent dictionary space through our proposed Wasserstein graph transformation (WGT). In WGT, we formulate inter-graph distance in W-metric space by virtue of the optimal transport (OT) principle, which effectively expresses the correlations of cross-graph structures. To make WGDL better representation ability, we dynamically update graph dictionary during training by maximizing the ratio of inter-class versus intra-class Wasserstein distance. To evaluate our WGDL method, comprehensive experiments are conducted on six graph classification datasets. Experimental results demonstrate the effectiveness of our WGDL, and state-of-the-art performance.

A Multi-Level Convolution Pyramid Semantic Fusion Framework for High-Resolution Remote Sensing Image Scene Classification and Annotation

High spatial resolution (HSR) imagery scene classification has become a hot research topic in remote sensing. Scene classification method based on the handcrafted features, such as the bag-of-visual-words (BoVW) model, describes an image by extracting local features of the scene and mapping them to the dictionary space, but usually uses a shallow structure and loses the spatial distribution characteristics of the scene. The method based on deep learning extracts hierarchical features to describe the scene, which can maintain the spatial position information well. However, deep features in different levels have scale recognition restrictions for multi-scale ground objects, and cannot understand complex scenes well. In this paper, the multi-level convolutional pyramid semantic fusion (MCPSF) framework is proposed for HSR imagery scene classification. Differing from previous scene classification methods, which integrate the feature of different levels directly, of which the fusion features have large differences in both sparsity and eigenvalue magnitude, MCPSF integrates multi-level semantic features extracted by BoVW model and convolutional neural network (CNN) model. In MCPSF, two convolution pyramid feature expression strategies are proposed to enhance the ability of capturing multi-scale land objects, i.e., local and convolutional pyramid based BoVW (LCPB) model and local and convolutional pyramid based pooling-stretched (LCPP) model. The effectiveness of the proposed method is verified on 21-class UC Merced (UCM) dataset and 30-class Aerial Image Dataset (AID). The framework was also transferred toa case study of scene annotation in Wuhan. The proposed framework significantly improves the performance when compared with other state-of-the-art methods.

Blind Quality Assessment of Screen Content Images Via Macro-Micro Modeling of Tensor Domain Dictionary

Screen content images (SCIs) have been rapidly and widely applied in interactive multimedia applications. The problem of quality assessment for SCIs is an interesting research topic. Most of the existing methods use subjective and independent features in gray domain to predict the image quality, which cannot comprehensively characterize the image properties or lack unified mathematical explanation for SCIs. To address these problems, we propose a novel blind quality assessment method based on macro-micro modeling of tensor domain dictionary for SCIs in this article. In the proposed method, the tensor decomposition is explored first to avoid the loss of color information, and then a target dictionary is learned more effectively with the principal components. Furthermore, a macro-micro model is established to characterize the micro and macro features in the target dictionary space, which can provide a systematic mathematical interpretation for feature extraction. For the micro features, a log-normal pooling scheme is designed to enhance the effectiveness of feature aggregation by analyzing the particularity of the statistical distribution of sparse codes. Additionally, the statistical properties are mainly discussed and studied based on the Bernoulli law of large numbers, and then a reliable macro feature is generated to describe the relationship between the statistical distribution and quality degradation of SCIs. Experimental results determined by using three public SCI databases show that the proposed method can perform better than relevant existing methods in the prediction of the visual quality of SCIs, especially in terms of the generalization for distortion type and interpretability for feature generation.

Classification of Alzheimer's Disease in MRI based on Dictionary Learning and Heavy Tailed Modelling.

Diagnosis of brain diseases is considered one of the most challenging medical tasks to perform, even for medical experts who rely on high-resolution anatomical images to identify signs of abnormalities by visual inspection. However, new computational tools which assist to automate this diagnosis have the potential to significantly improve the speed and accuracy of this process. This work presents a model to aid in the task of classification of structural Magnetic Resonance Imaging scans. The classification is performed using a Support Vector Machine, whilst the features to analyze belong to a dictionary space. Such space was mainly built from a dictionary learning perspective, although a predefined one was also assessed. The results indicate that features learnt from the data of interest lead to improved classification performance. The proposed framework was tested on the ADNI dataset stage I.

Dictionary Compression in Point Cloud Data Management

Nowadays, massive amounts of point cloud data can be collected thanks to advances in data acquisition and processing technologies such as dense image matching and airborne LiDAR scanning. With the increase in volume and precision, point cloud data offers a useful source of information for natural-resource management, urban planning, self-driving cars, and more. At the same time, on the scale that point cloud data is produced, management challenges are introduced: it is important to achieve efficiency both in terms of querying performance and space requirements. Traditional file-based solutions to point cloud management offer space efficiency, however, they cannot scale to such massive data and provide the declarative power of a DBMS. In this article, we propose a time- and space-efficient solution to storing and managing point cloud data in main memory column-store DBMS. Our solution, Space-Filling Curve Dictionary-Based Compression (SFC-DBC), employs dictionary-based compression in the spatial data management domain and enhances it with indexing capabilities by using space-filling curves. SFC-DBC does so by constructing the space-filling curve over a compressed, artificially introduced dictionary space. Consequently, SFC-DBC significantly optimizes query execution and yet does not require additional storage resources, compared to traditional dictionary-based compression. With respect to space-filling-curve-based approaches, it minimizes storage footprint and increases resilience to skew. As a proof of concept, we develop and evaluate our approach as a research prototype in the context of SAP HANA. SFC-DBC outperforms other dictionary-based compression schemes by up to 61% in terms of space and up to 9.4× in terms of query performance.

Unsupervised Scene Change Detection via Latent Dirichlet Allocation and Multivariate Alteration Detection

Scene change detection is the process of identifying the differences between the multitemporal image scenes, which has significant potential in the application of urban development and land management at the semantic level. Traditional scene change detection methods are based on the supervised scene classification, and then directly compare the independent classification results without considering the temporal correlation between the unchanged regions. However, few studies have focused on detecting the semantic changes of multitemporal image scenes with unsupervised methods. In this paper, we propose a novel unsupervised scene change detection method by using latent Dirichlet allocation (LDA) and multivariate alteration detection (MAD). First, the scene is represented by the bag-of-visual-words model, and adopt the union dictionary to ensure the consistency of dictionary space. Then, LDA is used to achieve the middle-level feature dimension reduction, and generate the common topic space of the two multitemporal image scene datasets. And finally, the MAD method was applied to detect the semantic changes of corresponding multitemporal image scenes. Two experiments with high-resolution remote sensing image scene datasets demonstrated that our proposed approach can get a better performance in unsupervised scene change detection without prior knowledge.

Speech enhancement in spectral envelope and details subspaces

Supervised speech enhancement techniques have been proved to improve speech intelligibility. However, one major challenge of supervised approaches involves the overlapped spectral bases between speech and noise components in spectral dictionary space. In this study, we address this challenge through a combination strategy of spectral modulation decoupling and low-rank and sparsity oriented decomposition. Specifically, supervised low-rank and sparse decompositions with energy thresholding are developed in the spectral envelop subspace, In the spectral details subspace, an unsupervised robust principal component analysis is utilized to extract the fine structure. The validation results show that, compared with five speech enhancement algorithms, including MMSE-SPP, NMF-RPCA, RPCA, LARC and BNMF, the proposed algorithms achieves satisfactory performance on improving both perceptual quality and speech intelligibility.

Joint Semantic and Latent Attribute Modelling for Cross-Class Transfer Learning.

A number of vision problems such as zero-shot learning and person re-identification can be considered as cross-class transfer learning problems. As mid-level semantic properties shared cross different object classes, attributes have been studied extensively for knowledge transfer across classes. Most previous attribute learning methods focus only on human-defined/nameable semantic attributes, whilst ignoring the fact there also exist undefined/latent shareable visual properties, or latent attributes. These latent attributes can be either discriminative or non-discriminative parts depending on whether they can contribute to an object recognition task. In this work, we argue that learning the latent attributes jointly with user-defined semantic attributes not only leads to better representation but also helps semantic attribute prediction. A novel dictionary learning model is proposed which decomposes the dictionary space into three parts corresponding to semantic, latent discriminative and latent background attributes respectively. Such a joint attribute learning model is then extended by following a multi-task transfer learning framework to address a more challenging unsupervised domain adaptation problem, where annotations are only available on an auxiliary dataset and the target dataset is completely unlabelled. Extensive experiments show that the proposed models, though being linear and thus extremely efficient to compute, produce state-of-the-art results on both zero-shot learning and person re-identification.

Learning Shape Representations for Multi-Atlas Endocardium Segmentation in 3D Echo Images

As part of the CETUS challenge, we present a multi-atlas segmentation framework to delineate the left-ventricle endocardium in echocardiographic images. To increase the robustness of the registration step, we introduce a speckle reduction step and a new shape representation based on sparse coding and manifold approximation in dictionary space. The shape representation, unlike intensity values, provides consistent shape information across different images. The validation results on the test set show that registration based on our shape representation significantly improves the performance of multi-atlas segmentation compared to intensity based registration. To our knowledge it is the first time that multi-atlas segmentation achieves state-of-the-art results for echocardiographic images.

Locality-Constrained Low-Rank Coding for Image Classification

Low-rank coding (LRC), originated from matrix decomposition, is recently introduced into image classification. Following the standard bag-of-words (BOW) pipeline, when coding the data matrix in the sense of low-rankness incorporates contextual information into the traditional BOW model, this can capture the dependency relationship among neighbor patches. It differs from the traditional sparse coding paradigms which encode patches independently. Current LRC-based methods use l_1 norm to increase the discrimination and sparseness of the learned codes. However, such methods fail to consider the local manifold structure between dataspace and dictionary space. To solve this problem, we propose a locality-constrained low-rank coding (LCLR) algorithm for image representations. By using the geometric structure information as a regularization term,we can obtain more discriminative representations. In addition, we present a fast and stable online algorithmto solve the optimization problem. In the experiments,we evaluate LCLR with four benchmarks, including one face recognition dataset (extended Yale B), one handwrittendigit recognition dataset (USPS), and two image datasets (Scene13 for scene recognition and Caltech101 for object recognition). Experimental results show thatour approach outperforms many state-of-the-art algorithmseven with a linear classifier.

Learning Ancestral Atom via Sparse Coding

Sparse signal models have been the focus of recent research. In sparse coding, signals are represented with a linear combination of a small number of elementary signals called atoms, and the collection of atoms is called a dictionary. Design of the dictionary has strong influence on the signal approximation performance. Recently, to put prior information into dictionary learning, several methods imposing a certain kind of structure on the dictionary are proposed. In this paper, like wavelet analysis, a dictionary for sparse signal representation is assumed to be generated from an ancestral atom, and a method for learning the ancestral atom is proposed. The proposed algorithm updates the ancestral atom by iterating dictionary update in unstructured dictionary space and projection of the updated dictionary onto the structured dictionary space. The algorithm allows a simple differential geometric interpretation. Numerical experiments are performed to show the characteristics and advantages of the proposed algorithm.